Unlocking the Latent Canvas: Eliciting and Benchmarking Symbolic Visual Expression in LLMs

TL;DR

This paper introduces SVE-ASCII, a framework for eliciting and benchmarking symbolic visual expression within Large Language Models using ASCII art, demonstrating that generative training improves visual understanding.

Contribution

It presents a novel ASCII art dataset, a unified instruction-tuning approach, and empirical evidence of the mutual reinforcement between visual perception and generation in LLMs.

Findings

Generative training enhances visual comprehension in LLMs.

The ASCIIArt-7K dataset enables systematic benchmarking.

A unified framework effectively elicits symbolic visual expression.

Abstract

Current multimodal approaches predominantly treat visual generation as an external process, relying on pixel rendering or code execution, thereby overlooking the native visual representation capabilities latent within Large Language Models (LLMs). In this work, we unlock this potential through ASCII art, a compact, efficient, and text-native visual format. We introduce SVE-ASCII, a unified framework designed to elicit and benchmark Symbolic Visual Expression directly within the pure text space. To address the scarcity of systematic resources, we construct ASCIIArt-7K, a high-quality dataset synthesized via a novel "Seed-and-Evolve" pipeline that augments human-curated anchors through in-context stylistic editing. We further implement a unified instruction-tuning strategy that jointly optimizes for both Generation (Text-to-ASCII) and Understanding (ASCII-to-Text). Crucially, our experiments reveal a critical phenomenon regarding task duality: while it is established that perception aids generation, we provide compelling evidence that generative training significantly enhances visual comprehension. This confirms a mutually reinforcing cycle in symbolic visual processing, a relationship previously hypothesized but rarely empirically demonstrated in the visual domain. We release our dataset, the ASCIIArt-Bench benchmark, and the SVE-ASCII model, establishing a robust baseline for native text-based visual intelligence.