Question Answering on HaluEval QA
45.4AccuracyHaluSearch (MCTS)
Evaluation Results
| Method | Links | |
|---|---|---|
| HaluSearch (MCTS)Backbone=Llama3.1-8B-Instruct2025.01 | 45.4 | |
| HaluSearch (MCTS)Backbone=Qwen2-7B-Instruct2025.01 | 43.69 | |
| BoNBackbone=Qwen2-7B-Instruct2025.01 | 39 | |
| BoNBackbone=Llama3.1-8B-Instruct2025.01 | 37.8 | |
| SCBackbone=Qwen2-7B-Instruct2025.01 | 37.2 | |
| CoTBackbone=Llama3.1-8B-Instruct2025.01 | 36.8 | |
| Self-RefineBackbone=Qwen2-7B-Instruct2025.01 | 35.8 | |
| Direct GenerationBackbone=Llama3.1-8B-Instruct2025.01 | 35.6 | |
| CoTBackbone=Qwen2-7B-Instruct2025.01 | 35.6 | |
| ITIBackbone=Llama3.1-8B-Instruct2025.01 | 35.2 | |
| SCBackbone=Llama3.1-8B-Instruct2025.01 | 35 | |
| Direct GenerationBackbone=Qwen2-7B-Instruct2025.01 | 34 | |
| ITIBackbone=Qwen2-7B-Instruct2025.01 | 32.2 | |
| Self-RefineBackbone=Llama3.1-8B-Instruct2025.01 | 28.2 |