MEDVQA-GI-2026-Task2-Explainability / metrics /t2_hybrid_metrics.json
sageofai's picture
t2_hybrid_metrics.json
86a3d3b verified
Raw
History Blame
252 Bytes
{
"Exact-Match": 0.038,
"Token-F1": 0.4456,
"BLEU-4": 0.1818,
"BERTScore-F1": 0.5342,
"Safety-Score": 0.8309,
"Halluc-Rate": 0.328,
"Overconf-Rate": 0.0,
"Abstention-Rate": 0.0,
"Mean-Confidence": 0.7759,
"ECE": 0.7359,
"N": 500
}