Information Coverage and Truthfulness Evaluation on Corpus-based Retrieval
34.4S_factMixtral-8x22B
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Mixtral-8x22BAlignment LLM=Llama-3.1-70B2025.01 | 34.4 | 37 | 29.7 | 41.4 | 34.2 | 40.9 | 33.9 | |
| GPT-4Alignment LLM=Llama-3.1-70B2025.01 | 34.3 | 41.6 | 32.7 | 45.3 | 34.6 | 46.3 | 35.4 | |
| Openchat 3.5 (7B)Alignment LLM=Llama-3.1-70B2025.01 | 34 | 41.3 | 32.9 | 42.9 | 34.8 | 42.4 | 34.7 | |
| Llama-3-70BAlignment LLM=Llama-3.1-70B2025.01 | 32.7 | 45.1 | 33.5 | 46.4 | 35.5 | 46.6 | 35.4 |