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

Failure prediction on SWE-agent (Transfer held-out)

0.69ROC-AUC

Src-aware

0.535040.575270.61550.65573Jun 22, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
0.69--
0.690.6230.001
0.6460.5730.001
2026.06
0.633--
2026.06
0.633--
2026.06
0.6240.5590.001
2026.06
0.541--