Style2Code: A Style-Controllable Code Generation Framework with Dual-Modal Contrastive Representation Learning

2025-05-26

AI Summary

Controllable code generation faces the challenge of simultaneously achieving precise stylistic control and preserving functional correctness. To address this, we propose a two-stage framework: (1) a dual-modal contrastive learning stage that jointly aligns code representations across stylistic, semantic, and structural dimensions; and (2) a conditional fine-tuning stage where Flan-T5 is adapted using learned style vectors to enable fine-grained, style-controllable generation. Our approach is the first to integrate contrastive representation alignment with conditional decoding, supporting style interpolation and lightweight user-specific personalization. Evaluated on multiple benchmarks, our method achieves significant improvements in style consistency (+12.3%) and editability, while maintaining high syntactic validity (98.6% correct syntax) and functional equivalence (94.1% behavior-preserving outputs), outperforming existing state-of-the-art methods.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Controllable code generation, the ability to synthesize code that follows a specified style while maintaining functionality, remains a challenging task. We propose a two-stage training framework combining contrastive learning and conditional decoding to enable flexible style control. The first stage aligns code style representations with semantic and structural features. In the second stage, we fine-tune a language model (e.g., Flan-T5) conditioned on the learned style vector to guide generation. Our method supports style interpolation and user personalization via lightweight mixing. Compared to prior work, our unified framework offers improved stylistic control without sacrificing code correctness. This is among the first approaches to combine contrastive alignment with conditional decoding for style-guided code generation.

Problem

• Enabling controllable code generation with specified styles
• Aligning code style with semantic and structural features
• Combining contrastive learning and conditional decoding

Innovation

• Dual-modal contrastive learning for style alignment
• Conditional decoding with fine-tuned Flan-T5
• Lightweight mixing for style interpolation