Question Answering on COQA (factual accuracy)
28.27Factual AccuracyARREST-HB
Evaluation Results
| Method | Links | |
|---|---|---|
| ARREST-HBLLM=Vicuna 7B2026.01 | 28.27 | |
| DOLALLM=Vicuna 7B2026.01 | 27.5 | |
| Activation DecodingLLM=Vicuna 7B2026.01 | 25.02 | |
| Self-ReflectionLLM=Vicuna 7B2026.01 | 23.69 | |
| ARREST-HBLLM=LLaMa-2 7B2026.01 | 21.59 | |
| ITILLM=LLaMa-2 7B2026.01 | 20.37 | |
| ARREST-HBLLM=LLaMa-3.1 8B2026.01 | 19.71 | |
| DOLALLM=LLaMa-2 7B2026.01 | 18.3 | |
| DOLALLM=LLaMa-3.1 8B2026.01 | 18.3 | |
| ITILLM=LLaMa-3.1 8B2026.01 | 16.54 | |
| Activation DecodingLLM=LLaMa-2 7B2026.01 | 16.3 | |
| Activation DecodingLLM=LLaMa-3.1 8B2026.01 | 16.3 | |
| Self-ReflectionLLM=LLaMa-2 7B2026.01 | 16 | |
| COVELLM=LLaMa-2 7B2026.01 | 15.5 | |
| Base ModelLLM=LLaMa-2 7B2026.01 | 15.1 | |
| COVELLM=Vicuna 7B2026.01 | 15.04 | |
| COVELLM=LLaMa-3.1 8B2026.01 | 15 | |
| ITILLM=Vicuna 7B2026.01 | 14.45 | |
| Base ModelLLM=Vicuna 7B2026.01 | 14.4 | |
| Self-ReflectionLLM=LLaMa-3.1 8B2026.01 | 13.69 | |
| Base ModelLLM=LLaMa-3.1 8B2026.01 | 13.27 |