Question Answering on TruthfulQA (Coverage, Efficiency, Entropy)
81CoverageCP
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CPBackbone LLM=Llama-2-7b-chat, Alpha (α)=0.2, Prompting technique=Chain-of-Thought, Number of samples=1002025.12 | 81 | 2.92 | 4 | |
| EC3Backbone LLM=Llama-2-7b-chat, Alpha (α)=0.2, Prompting technique=Chain-of-Thought, Number of samples=1002025.12 | 81 | 2.32 | 189 | |
| CONFTRBackbone LLM=Llama-2-7b-chat, Alpha (α)=0.2, Prompting technique=Chain-of-Thought, Number of samples=1002025.12 | 80 | 2.48 | 184 |