π backfill:2025-10-15..2025-10-15 Β· Multimodal Coding β 1 paper(s)
Auto-processed by the arXiv crawler pipeline. Review each paper and reply with the commands below.
1. RECODE: Reasoning Through Code Generation for Visual Question Answering
Authors: Junhong Shen, Mu Cai, Bo Hu, Ameet Talwalkar, David A Ross, et al.
Venue: arXiv 2025/10
RECODE is an agentic framework that performs visual reasoning by derendering images into executable code to enable verifiable symbolic inference. It iteratively generates, critiques, and refines code reconstructions of structured visuals like charts and diagrams to improve accuracy on complex multimodal tasks.
π Paper
Review commands (comment on this issue):
/approve all β accept all papers/approve 1,3 β accept papers 1 and 3/reject 2 β discard paper 2/approve 1,3 /reject 2 β mixed/edit 1 category=code_generation β change category before approving/edit 1 venue=ICSE 2026 β fix venue
π backfill:2025-10-15..2025-10-15 Β· Multimodal Coding β 1 paper(s)
1. RECODE: Reasoning Through Code Generation for Visual Question Answering
Authors: Junhong Shen, Mu Cai, Bo Hu, Ameet Talwalkar, David A Ross, et al.
Venue: arXiv 2025/10
Review commands (comment on this issue):
/approve allβ accept all papers/approve 1,3β accept papers 1 and 3/reject 2β discard paper 2/approve 1,3 /reject 2β mixed/edit 1 category=code_generationβ change category before approving/edit 1 venue=ICSE 2026β fix venue