Text2Arch is a new dataset of text descriptions, images, and DOT code for scientific architecture diagrams that enables fine-tuned models to generate diagrams outperforming baselines like DiagramAgent while matching GPT-4o.
## Inputs: IMAGE:#imageURL# Initial DOT code: (which may contain errors or incomplete data) #dotCode# 20 Published as a conference paper at ICLR 2026 (a)Original figure(Fig
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Text2Arch: A Dataset for Generating Scientific Architecture Diagrams from Natural Language Descriptions
Text2Arch is a new dataset of text descriptions, images, and DOT code for scientific architecture diagrams that enables fine-tuned models to generate diagrams outperforming baselines like DiagramAgent while matching GPT-4o.