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

Attribution Faithfulness Evaluation on IFEval, EvalPlus, MATH, and GAIA2 60 failure cases

1.7Average Tokens to Fix

AttnLRP

1.622.162.73.24Apr 20, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
1.7100
2026.04
2100
2026.04
3.796.7