ViCo: Visual-oriented Coding with Self-Reflection for Chart Replication

Researchers introduced ViCo, a training framework for visual-oriented coding that uses iterative self-reflection to progressively align generated chart images with reference designs, according to an arXiv paper (arXiv:2609.16014v1). ViCo combines a self-supervised warm-up stage that augments Monte Carlo Tree Search with consistency-based pruning to synthesize reflection trajectories, plus a multi-step reinforcement learning algorithm using counterfactual baselines to address reward sparsity. Trained on an 8B model, ViCo achieved performance close to proprietary LLMs with adequate reflection capabilities across three public benchmarks.

arXiv:2609.16014v1 Announce Type: new Abstract: This paper addresses the challenge of generating high-quality academic charts that match the visual standards of human-authored papers. While existing AI agents can produce well-structured text and code, their generated visualizations often lack the stylistic and semantic fidelity of human designs. Advanced coding agents that employ self-reflection mechanisms exhibit poor visual reasoning and limited reflection following, resulting in sparse reward signals that severely undermine their reinforcement learning RL . We propose ViCo, a training framework for visual-oriented coding that employs iterative reflections to align generated chart images progressively with the reference. We first introduce a self-supervised warm-up stage, which augments Monte Carlo Tree Search with consistency-based pruning to synthesize high-quality reflection trajectories, ensuring that each coding step strictly follows the outcomes of prior reflections. A multi-step RL algorithm is then developed, using counterfactual baselines to estimate advantage for reflection and action steps within each refinement cycle, thereby addressing the reward sparsity. To enable efficient reward in massive training, we propose an automatic, multifaceted evaluation framework that assesses charts' style, layout, and semantic consistency via a hierarchical heterogeneous layout graph structure. Experiments on three public benchmarks demonstrate that ViCo, trained on an 8B model, achieves performance close to proprietary LLMs with adequate reflection capabilities.