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

Multi-step Reasoning over Code Dependencies on ConvFinQA hard

74.67Accuracy (Multi-step Reasoning)

SGKR

51.093257.214163.33569.4559Apr 12, 2026
Updated 5d ago

Evaluation Results

MethodLinks
2026.04
74.67
2026.04
71.67
2026.04
71.33
2026.04
68.33
2026.04
63
2026.04
52