vince62s commited on
Commit
9b228ce
1 Parent(s): ffa30ae

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +66 -1
README.md CHANGED
@@ -31,4 +31,69 @@ Installation instruction:
31
  Visit: https://github.com/OpenNMT/OpenNMT-py
32
  make sure you install flash-attn and autoawq
33
 
34
- Enjoy
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
31
  Visit: https://github.com/OpenNMT/OpenNMT-py
32
  make sure you install flash-attn and autoawq
33
 
34
+ Enjoy
35
+
36
+ detailed MMLU scoring:
37
+ ```
38
+ ACC-abstract_algebra: 0.3600
39
+ ACC-anatomy: 0.6444
40
+ ACC-astronomy: 0.7303
41
+ ACC-business_ethics: 0.6400
42
+ ACC-clinical_knowledge: 0.7283
43
+ ACC-college_biology: 0.8056
44
+ ACC-college_chemistry: 0.5300
45
+ ACC-college_computer_science: 0.5900
46
+ ACC-college_mathematics: 0.3700
47
+ ACC-college_medicine: 0.6936
48
+ ACC-college_physics: 0.4510
49
+ ACC-computer_security: 0.7900
50
+ ACC-conceptual_physics: 0.6468
51
+ ACC-econometrics: 0.5614
52
+ ACC-electrical_engineering: 0.6414
53
+ ACC-elementary_mathematics: 0.4630
54
+ ACC-formal_logic: 0.4524
55
+ ACC-global_facts: 0.4600
56
+ ACC-high_school_biology: 0.8000
57
+ ACC-high_school_chemistry: 0.5320
58
+ ACC-high_school_computer_science: 0.7400
59
+ ACC-high_school_european_history: 0.8121
60
+ ACC-high_school_geography: 0.8081
61
+ ACC-high_school_government_and_politics: 0.9275
62
+ ACC-high_school_macroeconomics: 0.6923
63
+ ACC-high_school_mathematics: 0.3667
64
+ ACC-high_school_microeconomics: 0.7731
65
+ ACC-high_school_physics: 0.4636
66
+ ACC-high_school_psychology: 0.8569
67
+ ACC-high_school_statistics: 0.5278
68
+ ACC-high_school_us_history: 0.8431
69
+ ACC-high_school_world_history: 0.8650
70
+ ACC-human_aging: 0.7175
71
+ ACC-human_sexuality: 0.7710
72
+ ACC-international_law: 0.8347
73
+ ACC-jurisprudence: 0.7778
74
+ ACC-logical_fallacies: 0.7791
75
+ ACC-machine_learning: 0.5357
76
+ ACC-management: 0.7767
77
+ ACC-marketing: 0.9145
78
+ ACC-medical_genetics: 0.7100
79
+ ACC-miscellaneous: 0.8404
80
+ ACC-moral_disputes: 0.7775
81
+ ACC-moral_scenarios: 0.4112
82
+ ACC-nutrition: 0.7876
83
+ ACC-philosophy: 0.7492
84
+ ACC-prehistory: 0.7963
85
+ ACC-professional_accounting: 0.5177
86
+ ACC-professional_law: 0.5111
87
+ ACC-professional_medicine: 0.7390
88
+ ACC-professional_psychology: 0.7304
89
+ ACC-public_relations: 0.6727
90
+ ACC-security_studies: 0.7061
91
+ ACC-sociology: 0.8706
92
+ ACC-us_foreign_policy: 0.9100
93
+ ACC-virology: 0.5060
94
+ ACC-world_religions: 0.8538
95
+ ACC-all: 0.6707
96
+ [2023-12-22 16:35:03,999 INFO] total run time 7156.16
97
+
98
+ ```
99
+