English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The human brain is naturally equipped to comprehend and interpret visualinformation rapidly. When confronted with complex problems or concepts, we useflowcharts, sketches, and diagrams to aid our thought process. Leveraging thisinherent ability can significantly enhance logical reasoning. However, currentLarge Language Models (LLMs) do not utilize such visual intuition to help theirthinking. Even the most advanced version language models (e.g., GPT-4V andLLaVA) merely align images into textual space, which means their reasoningprocesses remain purely verbal. To mitigate such limitations, we present aChain of Images (CoI) approach, which can convert complex language reasoningproblems to simple pattern recognition by generating a series of images asintermediate representations. Furthermore, we have developed a CoI evaluationdataset encompassing 15 distinct domains where images can intuitively aidproblem-solving. Based on this dataset, we aim to construct a benchmark toassess the capability of future multimodal large-scale models to leverageimages for reasoning. In supporting our CoI reasoning, we introduce a symbolicmultimodal large language model (SyMLLM) that generates images strictly basedon language instructions and accepts both text and image as input. Experimentson Geometry, Chess and Common Sense tasks sourced from the CoI evaluationdataset show that CoI improves performance significantly over the pure-languageChain of Thoughts (CoT) baselines. The code is available athttps://github.com/GraphPKU/CoI.

Chain of Images for Intuitively Reasoning