Skip to yearly menu bar Skip to main content


Poster
in
Workshop: The 3rd DL4C Workshop: Emergent Possibilities and Challenges in Deep Learning for Code

CodeTransEngine: Ready-to-use Backend for LLM-based Code Translation

Marcos Macedo · Yuan Tian · Bram Adams


Abstract:

Code translation, the process of converting a program from one programming language to another, plays a significant role in modernizing legacy systems, ensuring cross-platform compatibility, and improving performance of programs. With the rise of Large Language Models (LLMs), new possibilities have emerged for automating this complex task. However, building effective LLM-based code translation pipelines involves numerous challenges, including prompt engineering, efficient inference, output control, test case validation, and resource optimization. To address these challenges, we introduce CodeTransEngine, a ready-to-use backend that simplifies LLM-based code translation. CodeTransEngine lowers the entry barrier for practitioners and reduces the effort required to leverage the power of LLMs for code translation, enabling them to focus on achieving their translation goals.

Chat is not available.