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

Multi-document Question Answering on MDQA (Accuracy Thresholds)

72.6Accuracy @ 0% Threshold

Attention Buckets

67.19268.5967071.404Jun 26, 2026
Updated 27d ago

Evaluation Results

MethodLinks
2026.06
72.661.460.660.859.663
2026.06
72.661.461.8625963.5
2026.06
71.661.260.660.862.463.3
2026.06
71.462.2626161.663.6
2026.06
71.259.658.856.456.260.4
2026.06
70.45855.455.460.459.9
2026.06
69.658.658.26065.462.4
2026.06
69.662.26360.26263.4
2026.06
69.664.869.26365.466.4
2026.06
69.46162.658.663.663
2026.06
69.461.863.460.261.463.2
2026.06
69.25959.859.267.463
2026.06
68.86060.86168.264.5
2026.06
68.66262.258.46463
2026.06
68.45760.26168.463
2026.06
68.461.2636163.863.5
2026.06
67.859.259.659.468.262.8
2026.06
67.46060.26068.663.2