List your company

Binary safety hallucination detection on PitVQA (External val)

96AccuracyQA-SNNE
13.8435.1756.577.83Nov 3, 2025

Evaluation results

MethodLinks
QA-SNNEBackbone Model=PitVQA, Evaluation Protocol=Zero-shot, Variant=Ent2025.1196
QA-SNNEBackbone Model=PitVQA, Evaluation Protocol=Zero-shot, Variant=CrossE2025.1196
VL-UBackbone Model=PitVQA, Evaluation Protocol=Zero-shot2025.1193
QA-SNNEBackbone Model=Llama3.2, Evaluation Protocol=Zero-shot, Variant=Emb2025.1192
SNNEBackbone Model=Llama3.2, Evaluation Protocol=Zero-shot2025.1191
QA-SNNEBackbone Model=SurgicalGPT, Evaluation Protocol=Zero-shot, Variant=Ent2025.1183
QA-SNNEBackbone Model=SurgicalGPT, Evaluation Protocol=Zero-shot, Variant=CrossE2025.1183
QA-SNNEBackbone Model=Llama3.2, Evaluation Protocol=Zero-shot, Variant=CrossE2025.1179
QA-SNNEBackbone Model=MedGemma3.0, Evaluation Protocol=Zero-shot, Variant=Emb2025.1177
QA-SNNEBackbone Model=Llama3.2, Evaluation Protocol=Zero-shot, Variant=Ent2025.1174
QA-SNNEBackbone Model=MedGemma3.0, Evaluation Protocol=Zero-shot, Variant=Ent2025.1174
QA-SNNEBackbone Model=MedGemma3.0, Evaluation Protocol=Zero-shot, Variant=CrossE2025.1174
VL-UBackbone Model=Llama3.2, Evaluation Protocol=Zero-shot2025.1173
VL-UBackbone Model=MedGemma3.0, Evaluation Protocol=Zero-shot2025.1173
QA-SNNEBackbone Model=SurgicalGPT, Evaluation Protocol=Zero-shot, Variant=Emb2025.1166
VL-UBackbone Model=SurgicalGPT, Evaluation Protocol=Zero-shot2025.1164
VL-UBackbone Model=Qwen2.5, Evaluation Protocol=Zero-shot2025.1154
SNNEBackbone Model=MedGemma3.0, Evaluation Protocol=Zero-shot2025.1148
SNNEBackbone Model=Qwen2.5, Evaluation Protocol=Zero-shot2025.1144
QA-SNNEBackbone Model=Qwen2.5, Evaluation Protocol=Zero-shot, Variant=Emb2025.1131
QA-SNNEBackbone Model=Qwen2.5, Evaluation Protocol=Zero-shot, Variant=Ent2025.1130
QA-SNNEBackbone Model=Qwen2.5, Evaluation Protocol=Zero-shot, Variant=CrossE2025.1129
QA-SNNEBackbone Model=PitVQA, Evaluation Protocol=Zero-shot, Variant=Emb2025.1128
SNNEBackbone Model=PitVQA, Evaluation Protocol=Zero-shot2025.1127
SNNEBackbone Model=SurgicalGPT, Evaluation Protocol=Zero-shot2025.1117