Claim Extraction on MMCE Intent-Critical
3.56Reference-Based ScoreMICE
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MICEModel=Gemini 2.0 Flash, Evaluation Protocol=LLM-based Evaluation2026.02 | 3.56 | 48 | 68 | 100 | 100 | |
| MICEModel=Gemini 2.0 Flash, Evaluation Protocol=Human Evaluation2026.02 | 3.31 | 81 | 97 | 92 | 99 | |
| MLLM with ICLModel=Gemini 2.0 Flash, Evaluation Protocol=Human Evaluation2026.02 | 2.6 | 76 | 96 | 83 | 99 | |
| MLLM with ICLModel=Gemini 2.0 Flash, Evaluation Protocol=LLM-based Evaluation2026.02 | 2.08 | 68 | 82 | 94 | 100 |