CodeFuse-13B: A Pretrained Multi-lingual Code Large Language Model

TL;DR

CodeFuse-13B is a multilingual code large language model supporting over 40 programming languages and Chinese prompts, demonstrating strong performance in code tasks and real-world deployment feedback.

Contribution

Introduces CodeFuse-13B, a pre-trained multilingual code LLM optimized for Chinese and English, with extensive training data and evaluation on industry benchmarks and real-world feedback.

Findings

Achieves 37.10% HumanEval pass@1 score, ranking among top multilingual code LLMs.

Outperforms other models in Chinese prompt scenarios.

Validated through real-world deployment and human feedback.

Abstract

Code Large Language Models (Code LLMs) have gained significant attention in the industry due to their wide applications in the full lifecycle of software engineering. However, the effectiveness of existing models in understanding non-English inputs for multi-lingual code-related tasks is still far from well studied. This paper introduces CodeFuse-13B, an open-sourced pre-trained code LLM. It is specifically designed for code-related tasks with both English and Chinese prompts and supports over 40 programming languages. CodeFuse achieves its effectiveness by utilizing a high quality pre-training dataset that is carefully filtered by program analyzers and optimized during the training process. Extensive experiments are conducted using real-world usage scenarios, the industry-standard benchmark HumanEval-x, and the specially designed CodeFuseEval for Chinese prompts. To assess the effectiveness of CodeFuse, we actively collected valuable human feedback from the AntGroup's software development process where CodeFuse has been successfully deployed. The results demonstrate that CodeFuse-13B achieves a HumanEval pass@1 score of 37.10%, positioning it as one of the top multi-lingual code LLMs with similar parameter sizes. In practical scenarios, such as code generation, code translation, code comments, and testcase generation, CodeFuse performs better than other models when confronted with Chinese prompts.