Hallucination Detection on PathVQA
82AUCCEBaG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CEBaGMLLM Backbone=HuatuoGPT-Vision-7B2026.03 | 82 | 55.6 | |
| CEBaGMLLM Backbone=LLaVA-Med-v1.5-Mistral-7B2026.03 | 77.6 | 53.1 | |
| AvgProbMLLM Backbone=LLaVA-Med-v1.5-Mistral-7B2026.03 | 75.1 | 52 | |
| CEBaGMLLM Backbone=MedGemma-4b-it2026.03 | 71.9 | 59 | |
| CEBaGMLLM Backbone=MedGemma-1.5-4b-it2026.03 | 66.6 | 48.8 | |
| AvgProbMLLM Backbone=MedGemma-1.5-4b-it2026.03 | 65.9 | 48.9 | |
| VASEMLLM Backbone=MedGemma-4b-it2026.03 | 61.5 | 49.6 | |
| VASEMLLM Backbone=LLaVA-Med-v1.5-Mistral-7B2026.03 | 60.4 | 41.7 | |
| VASEMLLM Backbone=MedGemma-1.5-4b-it2026.03 | 60.1 | 45.7 | |
| RadFlagMLLM Backbone=LLaVA-Med-v1.5-Mistral-7B2026.03 | 58.8 | 39.7 | |
| SEMLLM Backbone=MedGemma-4b-it2026.03 | 58 | 48.9 | |
| SEMLLM Backbone=LLaVA-Med-v1.5-Mistral-7B2026.03 | 57.4 | 41.3 | |
| RadFlagMLLM Backbone=MedGemma-4b-it2026.03 | 56.8 | 47.4 | |
| SEMLLM Backbone=MedGemma-1.5-4b-it2026.03 | 56.8 | 40.9 | |
| RadFlagMLLM Backbone=MedGemma-1.5-4b-it2026.03 | 56.3 | 41.6 | |
| RadFlagMLLM Backbone=HuatuoGPT-Vision-7B2026.03 | 55.5 | 38.9 | |
| SEMLLM Backbone=HuatuoGPT-Vision-7B2026.03 | 53.6 | 38.2 | |
| VASEMLLM Backbone=HuatuoGPT-Vision-7B2026.03 | 53.2 | 39.5 | |
| AvgProbMLLM Backbone=MedGemma-4b-it2026.03 | 46.4 | 45.1 | |
| AvgProbMLLM Backbone=HuatuoGPT-Vision-7B2026.03 | 30.5 | 28.7 |