Farouk commited on
Commit
a41002e
1 Parent(s): 28f09c4

commit files to HF hub

Browse files
Files changed (5) hide show
  1. adapter_model.bin +1 -1
  2. optimizer.pt +1 -1
  3. rng_state.pth +1 -1
  4. scheduler.pt +1 -1
  5. trainer_state.json +769 -5
adapter_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c7f87c0a8bee8578f5d38d3c4880e9dab0aef33b05a4732f6791102e2f143648
3
  size 319977229
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5865b0ec8c9cc8c34bd843204868211b91f67c6e0a8fa3b9dd9ed248967be1b
3
  size 319977229
optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:626f197c784017802ea98b2437c31d63211562db1f8737368c709cda1c5a1238
3
  size 1279539973
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c6de487f5037149f44bfbb41ed0efb84e91df86a60bca0cc1e3bd0244870b2fd
3
  size 1279539973
rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d898f6cda36de4bfc126de081edd9dac4866bd24da993c43c1deb6f43b8d016a
3
  size 14511
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:451e0285a66e29ba3a68db67fc317d748c9e822de6e4f95d124bb8fe74c8b1b1
3
  size 14511
scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:de7840bcb72f2f480fd301578d289cdfa174589e831b0d33e5772f3956b6beae
3
  size 627
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fcd2587f362188ac4728d4fa6edf8d2b0b6d72db365d49f7b847d4d79e3da09f
3
  size 627
trainer_state.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
- "best_metric": 0.8907580375671387,
3
- "best_model_checkpoint": "./output_v2/7b_cluster04_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_04/checkpoint-800",
4
- "epoch": 0.7238181406921511,
5
- "global_step": 800,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
@@ -770,11 +770,775 @@
770
  "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
771
  "mmlu_loss": 0.9363480592657163,
772
  "step": 800
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
773
  }
774
  ],
775
  "max_steps": 5000,
776
  "num_train_epochs": 5,
777
- "total_flos": 1.817341761037271e+17,
778
  "trial_name": null,
779
  "trial_params": null
780
  }
 
1
  {
2
+ "best_metric": 0.8871135711669922,
3
+ "best_model_checkpoint": "./output_v2/7b_cluster04_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_04/checkpoint-1000",
4
+ "epoch": 1.4476362813843022,
5
+ "global_step": 1600,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
 
770
  "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
771
  "mmlu_loss": 0.9363480592657163,
772
  "step": 800
773
+ },
774
+ {
775
+ "epoch": 0.73,
776
+ "learning_rate": 0.0002,
777
+ "loss": 0.8608,
778
+ "step": 810
779
+ },
780
+ {
781
+ "epoch": 0.74,
782
+ "learning_rate": 0.0002,
783
+ "loss": 0.8207,
784
+ "step": 820
785
+ },
786
+ {
787
+ "epoch": 0.75,
788
+ "learning_rate": 0.0002,
789
+ "loss": 0.8284,
790
+ "step": 830
791
+ },
792
+ {
793
+ "epoch": 0.76,
794
+ "learning_rate": 0.0002,
795
+ "loss": 0.8135,
796
+ "step": 840
797
+ },
798
+ {
799
+ "epoch": 0.77,
800
+ "learning_rate": 0.0002,
801
+ "loss": 0.8523,
802
+ "step": 850
803
+ },
804
+ {
805
+ "epoch": 0.78,
806
+ "learning_rate": 0.0002,
807
+ "loss": 0.8349,
808
+ "step": 860
809
+ },
810
+ {
811
+ "epoch": 0.79,
812
+ "learning_rate": 0.0002,
813
+ "loss": 0.855,
814
+ "step": 870
815
+ },
816
+ {
817
+ "epoch": 0.8,
818
+ "learning_rate": 0.0002,
819
+ "loss": 0.8376,
820
+ "step": 880
821
+ },
822
+ {
823
+ "epoch": 0.81,
824
+ "learning_rate": 0.0002,
825
+ "loss": 0.8906,
826
+ "step": 890
827
+ },
828
+ {
829
+ "epoch": 0.81,
830
+ "learning_rate": 0.0002,
831
+ "loss": 0.8522,
832
+ "step": 900
833
+ },
834
+ {
835
+ "epoch": 0.82,
836
+ "learning_rate": 0.0002,
837
+ "loss": 0.8491,
838
+ "step": 910
839
+ },
840
+ {
841
+ "epoch": 0.83,
842
+ "learning_rate": 0.0002,
843
+ "loss": 0.8421,
844
+ "step": 920
845
+ },
846
+ {
847
+ "epoch": 0.84,
848
+ "learning_rate": 0.0002,
849
+ "loss": 0.8285,
850
+ "step": 930
851
+ },
852
+ {
853
+ "epoch": 0.85,
854
+ "learning_rate": 0.0002,
855
+ "loss": 0.8655,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 0.86,
860
+ "learning_rate": 0.0002,
861
+ "loss": 0.8348,
862
+ "step": 950
863
+ },
864
+ {
865
+ "epoch": 0.87,
866
+ "learning_rate": 0.0002,
867
+ "loss": 0.8695,
868
+ "step": 960
869
+ },
870
+ {
871
+ "epoch": 0.88,
872
+ "learning_rate": 0.0002,
873
+ "loss": 0.8738,
874
+ "step": 970
875
+ },
876
+ {
877
+ "epoch": 0.89,
878
+ "learning_rate": 0.0002,
879
+ "loss": 0.8576,
880
+ "step": 980
881
+ },
882
+ {
883
+ "epoch": 0.9,
884
+ "learning_rate": 0.0002,
885
+ "loss": 0.7884,
886
+ "step": 990
887
+ },
888
+ {
889
+ "epoch": 0.9,
890
+ "learning_rate": 0.0002,
891
+ "loss": 0.8172,
892
+ "step": 1000
893
+ },
894
+ {
895
+ "epoch": 0.9,
896
+ "eval_loss": 0.8871135711669922,
897
+ "eval_runtime": 191.669,
898
+ "eval_samples_per_second": 5.217,
899
+ "eval_steps_per_second": 2.609,
900
+ "step": 1000
901
+ },
902
+ {
903
+ "epoch": 0.9,
904
+ "mmlu_eval_accuracy": 0.47325503865005797,
905
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
906
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
907
+ "mmlu_eval_accuracy_astronomy": 0.5,
908
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
909
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
910
+ "mmlu_eval_accuracy_college_biology": 0.4375,
911
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
912
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
913
+ "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
914
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
915
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
916
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
917
+ "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
918
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
919
+ "mmlu_eval_accuracy_electrical_engineering": 0.3125,
920
+ "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
921
+ "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
922
+ "mmlu_eval_accuracy_global_facts": 0.5,
923
+ "mmlu_eval_accuracy_high_school_biology": 0.34375,
924
+ "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
925
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
926
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
927
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
928
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
929
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
930
+ "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
931
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
932
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
933
+ "mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
934
+ "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
935
+ "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
936
+ "mmlu_eval_accuracy_high_school_world_history": 0.5,
937
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
938
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
939
+ "mmlu_eval_accuracy_international_law": 0.7692307692307693,
940
+ "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
941
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
942
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
943
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
944
+ "mmlu_eval_accuracy_marketing": 0.8,
945
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
946
+ "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
947
+ "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
948
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
949
+ "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
950
+ "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
951
+ "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
952
+ "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
953
+ "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
954
+ "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
955
+ "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
956
+ "mmlu_eval_accuracy_public_relations": 0.5,
957
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
958
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
959
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
960
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
961
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
962
+ "mmlu_loss": 0.9558495128084723,
963
+ "step": 1000
964
+ },
965
+ {
966
+ "epoch": 0.91,
967
+ "learning_rate": 0.0002,
968
+ "loss": 0.8471,
969
+ "step": 1010
970
+ },
971
+ {
972
+ "epoch": 0.92,
973
+ "learning_rate": 0.0002,
974
+ "loss": 0.8509,
975
+ "step": 1020
976
+ },
977
+ {
978
+ "epoch": 0.93,
979
+ "learning_rate": 0.0002,
980
+ "loss": 0.8344,
981
+ "step": 1030
982
+ },
983
+ {
984
+ "epoch": 0.94,
985
+ "learning_rate": 0.0002,
986
+ "loss": 0.8377,
987
+ "step": 1040
988
+ },
989
+ {
990
+ "epoch": 0.95,
991
+ "learning_rate": 0.0002,
992
+ "loss": 0.8533,
993
+ "step": 1050
994
+ },
995
+ {
996
+ "epoch": 0.96,
997
+ "learning_rate": 0.0002,
998
+ "loss": 0.8383,
999
+ "step": 1060
1000
+ },
1001
+ {
1002
+ "epoch": 0.97,
1003
+ "learning_rate": 0.0002,
1004
+ "loss": 0.8115,
1005
+ "step": 1070
1006
+ },
1007
+ {
1008
+ "epoch": 0.98,
1009
+ "learning_rate": 0.0002,
1010
+ "loss": 0.8691,
1011
+ "step": 1080
1012
+ },
1013
+ {
1014
+ "epoch": 0.99,
1015
+ "learning_rate": 0.0002,
1016
+ "loss": 0.8658,
1017
+ "step": 1090
1018
+ },
1019
+ {
1020
+ "epoch": 1.0,
1021
+ "learning_rate": 0.0002,
1022
+ "loss": 0.8226,
1023
+ "step": 1100
1024
+ },
1025
+ {
1026
+ "epoch": 1.0,
1027
+ "learning_rate": 0.0002,
1028
+ "loss": 0.8028,
1029
+ "step": 1110
1030
+ },
1031
+ {
1032
+ "epoch": 1.01,
1033
+ "learning_rate": 0.0002,
1034
+ "loss": 0.7684,
1035
+ "step": 1120
1036
+ },
1037
+ {
1038
+ "epoch": 1.02,
1039
+ "learning_rate": 0.0002,
1040
+ "loss": 0.8027,
1041
+ "step": 1130
1042
+ },
1043
+ {
1044
+ "epoch": 1.03,
1045
+ "learning_rate": 0.0002,
1046
+ "loss": 0.7238,
1047
+ "step": 1140
1048
+ },
1049
+ {
1050
+ "epoch": 1.04,
1051
+ "learning_rate": 0.0002,
1052
+ "loss": 0.8109,
1053
+ "step": 1150
1054
+ },
1055
+ {
1056
+ "epoch": 1.05,
1057
+ "learning_rate": 0.0002,
1058
+ "loss": 0.7647,
1059
+ "step": 1160
1060
+ },
1061
+ {
1062
+ "epoch": 1.06,
1063
+ "learning_rate": 0.0002,
1064
+ "loss": 0.8232,
1065
+ "step": 1170
1066
+ },
1067
+ {
1068
+ "epoch": 1.07,
1069
+ "learning_rate": 0.0002,
1070
+ "loss": 0.8071,
1071
+ "step": 1180
1072
+ },
1073
+ {
1074
+ "epoch": 1.08,
1075
+ "learning_rate": 0.0002,
1076
+ "loss": 0.7689,
1077
+ "step": 1190
1078
+ },
1079
+ {
1080
+ "epoch": 1.09,
1081
+ "learning_rate": 0.0002,
1082
+ "loss": 0.7889,
1083
+ "step": 1200
1084
+ },
1085
+ {
1086
+ "epoch": 1.09,
1087
+ "eval_loss": 0.8899393677711487,
1088
+ "eval_runtime": 191.3752,
1089
+ "eval_samples_per_second": 5.225,
1090
+ "eval_steps_per_second": 2.613,
1091
+ "step": 1200
1092
+ },
1093
+ {
1094
+ "epoch": 1.09,
1095
+ "mmlu_eval_accuracy": 0.479059525275743,
1096
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
1097
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
1098
+ "mmlu_eval_accuracy_astronomy": 0.5,
1099
+ "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
1100
+ "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
1101
+ "mmlu_eval_accuracy_college_biology": 0.375,
1102
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
1103
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
1104
+ "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
1105
+ "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
1106
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
1107
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
1108
+ "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
1109
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
1110
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
1111
+ "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
1112
+ "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
1113
+ "mmlu_eval_accuracy_global_facts": 0.5,
1114
+ "mmlu_eval_accuracy_high_school_biology": 0.375,
1115
+ "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
1116
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
1117
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
1118
+ "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
1119
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
1120
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
1121
+ "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
1122
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
1123
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
1124
+ "mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333,
1125
+ "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
1126
+ "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
1127
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
1128
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
1129
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
1130
+ "mmlu_eval_accuracy_international_law": 0.7692307692307693,
1131
+ "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
1132
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
1133
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
1134
+ "mmlu_eval_accuracy_management": 0.8181818181818182,
1135
+ "mmlu_eval_accuracy_marketing": 0.8,
1136
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
1137
+ "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
1138
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
1139
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
1140
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
1141
+ "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
1142
+ "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
1143
+ "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
1144
+ "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
1145
+ "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
1146
+ "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
1147
+ "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
1148
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
1149
+ "mmlu_eval_accuracy_sociology": 0.5909090909090909,
1150
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
1151
+ "mmlu_eval_accuracy_virology": 0.5,
1152
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
1153
+ "mmlu_loss": 1.0280111437425135,
1154
+ "step": 1200
1155
+ },
1156
+ {
1157
+ "epoch": 1.09,
1158
+ "learning_rate": 0.0002,
1159
+ "loss": 0.7596,
1160
+ "step": 1210
1161
+ },
1162
+ {
1163
+ "epoch": 1.1,
1164
+ "learning_rate": 0.0002,
1165
+ "loss": 0.768,
1166
+ "step": 1220
1167
+ },
1168
+ {
1169
+ "epoch": 1.11,
1170
+ "learning_rate": 0.0002,
1171
+ "loss": 0.7645,
1172
+ "step": 1230
1173
+ },
1174
+ {
1175
+ "epoch": 1.12,
1176
+ "learning_rate": 0.0002,
1177
+ "loss": 0.75,
1178
+ "step": 1240
1179
+ },
1180
+ {
1181
+ "epoch": 1.13,
1182
+ "learning_rate": 0.0002,
1183
+ "loss": 0.8108,
1184
+ "step": 1250
1185
+ },
1186
+ {
1187
+ "epoch": 1.14,
1188
+ "learning_rate": 0.0002,
1189
+ "loss": 0.7568,
1190
+ "step": 1260
1191
+ },
1192
+ {
1193
+ "epoch": 1.15,
1194
+ "learning_rate": 0.0002,
1195
+ "loss": 0.766,
1196
+ "step": 1270
1197
+ },
1198
+ {
1199
+ "epoch": 1.16,
1200
+ "learning_rate": 0.0002,
1201
+ "loss": 0.7783,
1202
+ "step": 1280
1203
+ },
1204
+ {
1205
+ "epoch": 1.17,
1206
+ "learning_rate": 0.0002,
1207
+ "loss": 0.767,
1208
+ "step": 1290
1209
+ },
1210
+ {
1211
+ "epoch": 1.18,
1212
+ "learning_rate": 0.0002,
1213
+ "loss": 0.7344,
1214
+ "step": 1300
1215
+ },
1216
+ {
1217
+ "epoch": 1.19,
1218
+ "learning_rate": 0.0002,
1219
+ "loss": 0.7976,
1220
+ "step": 1310
1221
+ },
1222
+ {
1223
+ "epoch": 1.19,
1224
+ "learning_rate": 0.0002,
1225
+ "loss": 0.7772,
1226
+ "step": 1320
1227
+ },
1228
+ {
1229
+ "epoch": 1.2,
1230
+ "learning_rate": 0.0002,
1231
+ "loss": 0.7825,
1232
+ "step": 1330
1233
+ },
1234
+ {
1235
+ "epoch": 1.21,
1236
+ "learning_rate": 0.0002,
1237
+ "loss": 0.8098,
1238
+ "step": 1340
1239
+ },
1240
+ {
1241
+ "epoch": 1.22,
1242
+ "learning_rate": 0.0002,
1243
+ "loss": 0.7797,
1244
+ "step": 1350
1245
+ },
1246
+ {
1247
+ "epoch": 1.23,
1248
+ "learning_rate": 0.0002,
1249
+ "loss": 0.7436,
1250
+ "step": 1360
1251
+ },
1252
+ {
1253
+ "epoch": 1.24,
1254
+ "learning_rate": 0.0002,
1255
+ "loss": 0.8052,
1256
+ "step": 1370
1257
+ },
1258
+ {
1259
+ "epoch": 1.25,
1260
+ "learning_rate": 0.0002,
1261
+ "loss": 0.7533,
1262
+ "step": 1380
1263
+ },
1264
+ {
1265
+ "epoch": 1.26,
1266
+ "learning_rate": 0.0002,
1267
+ "loss": 0.7764,
1268
+ "step": 1390
1269
+ },
1270
+ {
1271
+ "epoch": 1.27,
1272
+ "learning_rate": 0.0002,
1273
+ "loss": 0.8187,
1274
+ "step": 1400
1275
+ },
1276
+ {
1277
+ "epoch": 1.27,
1278
+ "eval_loss": 0.8930786848068237,
1279
+ "eval_runtime": 191.2133,
1280
+ "eval_samples_per_second": 5.23,
1281
+ "eval_steps_per_second": 2.615,
1282
+ "step": 1400
1283
+ },
1284
+ {
1285
+ "epoch": 1.27,
1286
+ "mmlu_eval_accuracy": 0.47881124581317025,
1287
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
1288
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
1289
+ "mmlu_eval_accuracy_astronomy": 0.5,
1290
+ "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
1291
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
1292
+ "mmlu_eval_accuracy_college_biology": 0.4375,
1293
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
1294
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
1295
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
1296
+ "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
1297
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
1298
+ "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
1299
+ "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
1300
+ "mmlu_eval_accuracy_econometrics": 0.25,
1301
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
1302
+ "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
1303
+ "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
1304
+ "mmlu_eval_accuracy_global_facts": 0.5,
1305
+ "mmlu_eval_accuracy_high_school_biology": 0.34375,
1306
+ "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
1307
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
1308
+ "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
1309
+ "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
1310
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
1311
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
1312
+ "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
1313
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
1314
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
1315
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
1316
+ "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
1317
+ "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
1318
+ "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
1319
+ "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
1320
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
1321
+ "mmlu_eval_accuracy_international_law": 0.7692307692307693,
1322
+ "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
1323
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
1324
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
1325
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
1326
+ "mmlu_eval_accuracy_marketing": 0.8,
1327
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
1328
+ "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
1329
+ "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
1330
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
1331
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
1332
+ "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
1333
+ "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
1334
+ "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
1335
+ "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
1336
+ "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
1337
+ "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
1338
+ "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
1339
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
1340
+ "mmlu_eval_accuracy_sociology": 0.7727272727272727,
1341
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
1342
+ "mmlu_eval_accuracy_virology": 0.5,
1343
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
1344
+ "mmlu_loss": 0.9709572936493329,
1345
+ "step": 1400
1346
+ },
1347
+ {
1348
+ "epoch": 1.28,
1349
+ "learning_rate": 0.0002,
1350
+ "loss": 0.7552,
1351
+ "step": 1410
1352
+ },
1353
+ {
1354
+ "epoch": 1.28,
1355
+ "learning_rate": 0.0002,
1356
+ "loss": 0.7832,
1357
+ "step": 1420
1358
+ },
1359
+ {
1360
+ "epoch": 1.29,
1361
+ "learning_rate": 0.0002,
1362
+ "loss": 0.7488,
1363
+ "step": 1430
1364
+ },
1365
+ {
1366
+ "epoch": 1.3,
1367
+ "learning_rate": 0.0002,
1368
+ "loss": 0.7568,
1369
+ "step": 1440
1370
+ },
1371
+ {
1372
+ "epoch": 1.31,
1373
+ "learning_rate": 0.0002,
1374
+ "loss": 0.7055,
1375
+ "step": 1450
1376
+ },
1377
+ {
1378
+ "epoch": 1.32,
1379
+ "learning_rate": 0.0002,
1380
+ "loss": 0.7729,
1381
+ "step": 1460
1382
+ },
1383
+ {
1384
+ "epoch": 1.33,
1385
+ "learning_rate": 0.0002,
1386
+ "loss": 0.7549,
1387
+ "step": 1470
1388
+ },
1389
+ {
1390
+ "epoch": 1.34,
1391
+ "learning_rate": 0.0002,
1392
+ "loss": 0.7508,
1393
+ "step": 1480
1394
+ },
1395
+ {
1396
+ "epoch": 1.35,
1397
+ "learning_rate": 0.0002,
1398
+ "loss": 0.7594,
1399
+ "step": 1490
1400
+ },
1401
+ {
1402
+ "epoch": 1.36,
1403
+ "learning_rate": 0.0002,
1404
+ "loss": 0.7596,
1405
+ "step": 1500
1406
+ },
1407
+ {
1408
+ "epoch": 1.37,
1409
+ "learning_rate": 0.0002,
1410
+ "loss": 0.801,
1411
+ "step": 1510
1412
+ },
1413
+ {
1414
+ "epoch": 1.38,
1415
+ "learning_rate": 0.0002,
1416
+ "loss": 0.7232,
1417
+ "step": 1520
1418
+ },
1419
+ {
1420
+ "epoch": 1.38,
1421
+ "learning_rate": 0.0002,
1422
+ "loss": 0.7736,
1423
+ "step": 1530
1424
+ },
1425
+ {
1426
+ "epoch": 1.39,
1427
+ "learning_rate": 0.0002,
1428
+ "loss": 0.7853,
1429
+ "step": 1540
1430
+ },
1431
+ {
1432
+ "epoch": 1.4,
1433
+ "learning_rate": 0.0002,
1434
+ "loss": 0.7783,
1435
+ "step": 1550
1436
+ },
1437
+ {
1438
+ "epoch": 1.41,
1439
+ "learning_rate": 0.0002,
1440
+ "loss": 0.7791,
1441
+ "step": 1560
1442
+ },
1443
+ {
1444
+ "epoch": 1.42,
1445
+ "learning_rate": 0.0002,
1446
+ "loss": 0.7517,
1447
+ "step": 1570
1448
+ },
1449
+ {
1450
+ "epoch": 1.43,
1451
+ "learning_rate": 0.0002,
1452
+ "loss": 0.7666,
1453
+ "step": 1580
1454
+ },
1455
+ {
1456
+ "epoch": 1.44,
1457
+ "learning_rate": 0.0002,
1458
+ "loss": 0.7825,
1459
+ "step": 1590
1460
+ },
1461
+ {
1462
+ "epoch": 1.45,
1463
+ "learning_rate": 0.0002,
1464
+ "loss": 0.7583,
1465
+ "step": 1600
1466
+ },
1467
+ {
1468
+ "epoch": 1.45,
1469
+ "eval_loss": 0.8902609944343567,
1470
+ "eval_runtime": 191.2523,
1471
+ "eval_samples_per_second": 5.229,
1472
+ "eval_steps_per_second": 2.614,
1473
+ "step": 1600
1474
+ },
1475
+ {
1476
+ "epoch": 1.45,
1477
+ "mmlu_eval_accuracy": 0.4726621848117947,
1478
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
1479
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
1480
+ "mmlu_eval_accuracy_astronomy": 0.375,
1481
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
1482
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
1483
+ "mmlu_eval_accuracy_college_biology": 0.4375,
1484
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
1485
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
1486
+ "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
1487
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
1488
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
1489
+ "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
1490
+ "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
1491
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
1492
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
1493
+ "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
1494
+ "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
1495
+ "mmlu_eval_accuracy_global_facts": 0.5,
1496
+ "mmlu_eval_accuracy_high_school_biology": 0.375,
1497
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
1498
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
1499
+ "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
1500
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
1501
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
1502
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
1503
+ "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
1504
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
1505
+ "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
1506
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
1507
+ "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
1508
+ "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
1509
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
1510
+ "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
1511
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
1512
+ "mmlu_eval_accuracy_international_law": 0.6923076923076923,
1513
+ "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
1514
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
1515
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
1516
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
1517
+ "mmlu_eval_accuracy_marketing": 0.8,
1518
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
1519
+ "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
1520
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
1521
+ "mmlu_eval_accuracy_moral_scenarios": 0.23,
1522
+ "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
1523
+ "mmlu_eval_accuracy_philosophy": 0.38235294117647056,
1524
+ "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
1525
+ "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
1526
+ "mmlu_eval_accuracy_professional_law": 0.3176470588235294,
1527
+ "mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
1528
+ "mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
1529
+ "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
1530
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
1531
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
1532
+ "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
1533
+ "mmlu_eval_accuracy_virology": 0.5,
1534
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
1535
+ "mmlu_loss": 0.8883641367637146,
1536
+ "step": 1600
1537
  }
1538
  ],
1539
  "max_steps": 5000,
1540
  "num_train_epochs": 5,
1541
+ "total_flos": 3.632227588441866e+17,
1542
  "trial_name": null,
1543
  "trial_params": null
1544
  }