MEDVQA-GI-2026-Task2-Explainability / metrics /t2_qwen_ft_metrics.json
sageofai's picture
t2_qwen_ft_metrics.json
f59ecde verified
Raw
History Blame
253 Bytes
{
"Exact-Match": 0.04,
"Token-F1": 0.4967,
"BLEU-4": 0.2025,
"BERTScore-F1": 0.5817,
"Safety-Score": 0.8511,
"Halluc-Rate": 0.368,
"Overconf-Rate": 0.0,
"Abstention-Rate": 0.002,
"Mean-Confidence": 0.7258,
"ECE": 0.6838,
"N": 500
}