Multi-task Language Understanding on MMLU (ACC, F1)
92MMLU AccuracyDIRECT QUERY
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DIRECT QUERYPRIVATE=NO, Backbone=GPT-4O2023.12 | 92 | 92 | |
| DIRECT QUERYPRIVATE=NO, Backbone=GPT-4-TURBO2023.12 | 89 | 71 | |
| CONFUSIONPROMPTPRIVATE=YES, Backbone=GPT-4O2023.12 | 89 | 90 | |
| CONFUSIONPROMPTPRIVATE=YES, Backbone=GPT-4-TURBO2023.12 | 83 | 82 | |
| DIRECT QUERYPRIVATE=NO, Backbone=GPT-3.5-TURBO2023.12 | 74 | 74 | |
| CONFUSIONPROMPTPRIVATE=YES, Backbone=GPT-3.5-TURBO2023.12 | 71 | 71 | |
| VICUNA-13BPRIVATE=YES2023.12 | 60 | 60 | |
| PARAPHRASERPRIVATE=YES, Backbone=GPT-4O2023.12 | 59 | 58 | |
| PARAPHRASERPRIVATE=YES, Backbone=GPT-4-TURBO2023.12 | 58 | 57 | |
| LLAMA2-7BPRIVATE=YES2023.12 | 56 | 54 | |
| TEXT2TEXTPRIVATE=YES, Backbone=GPT-4O2023.12 | 48 | 47 | |
| PARAPHRASERPRIVATE=YES, Backbone=GPT-3.5-TURBO2023.12 | 47 | 46 | |
| TEXT2TEXTPRIVATE=YES, Backbone=GPT-4-TURBO2023.12 | 45 | 45 | |
| TEXT2TEXTPRIVATE=YES, Backbone=GPT-3.5-TURBO2023.12 | 37 | 36 |