MFTCoder: Boosting Code LLMs with Multitask Fine-Tuning

TL;DR

MFTCoder introduces a multi-task fine-tuning framework for code LLMs that enhances performance, efficiency, and versatility by enabling simultaneous training on multiple tasks, outperforming single-task approaches and surpassing GPT-4 on benchmarks.

Contribution

The paper presents a novel multi-task fine-tuning method for code LLMs that improves performance and training efficiency, leveraging various loss functions and integrating with open-source models.

Findings

Outperforms single-task fine-tuning and ensemble methods.

Achieves 74.4% pass@1 on HumaneEval, surpassing GPT-4.

Offers efficient training with tokenization modes and PEFT.

Abstract

Code LLMs have emerged as a specialized research field, with remarkable studies dedicated to enhancing model's coding capabilities through fine-tuning on pre-trained models. Previous fine-tuning approaches were typically tailored to specific downstream tasks or scenarios, which meant separate fine-tuning for each task, requiring extensive training resources and posing challenges in terms of deployment and maintenance. Furthermore, these approaches failed to leverage the inherent interconnectedness among different code-related tasks. To overcome these limitations, we present a multi-task fine-tuning framework, MFTcoder, that enables simultaneous and parallel fine-tuning on multiple tasks. By incorporating various loss functions, we effectively address common challenges in multi-task learning, such as data imbalance, varying difficulty levels, and inconsistent convergence speeds. Extensive experiments have conclusively demonstrated that our multi-task fine-tuning approach outperforms both individual fine-tuning on single tasks and fine-tuning on a mixed ensemble of tasks. Moreover, MFTcoder offers efficient training capabilities, including efficient data tokenization modes and PEFT fine-tuning, resulting in significantly improved speed compared to traditional fine-tuning methods. MFTcoder seamlessly integrates with several mainstream open-source LLMs, such as CodeLLama and Qwen. Leveraging the CodeLLama foundation, our MFTcoder fine-tuned model, \textsc{CodeFuse-CodeLLama-34B}, achieves an impressive pass@1 score of 74.4\% on the HumaneEval benchmark, surpassing GPT-4 performance (67\%, zero-shot). MFTCoder is open-sourced at \url{https://github.com/codefuse-ai/MFTCOder}