Share your thoughts, 1 month free Claude Pro on usSee more

Multiple Choice QA on PubMedQA (test)

81.8AUROC

Total variance, experts

Updated 16d ago

Evaluation Results

Method	Links
Total variance, experts 2026.04		81.8
Total variance, CAE 2026.04		73.2
Total variance, base 2026.04		70.9
Entropy and MI 2026.04		70.7
Prediction variance 2026.04		70.2
Single model 2026.04		69.3
MC Dropout 2026.04		67.1
Test-time augmentations 2026.04		67
HUQ 2026.04		66.2