PanGu-Coder2: Boosting Large Language Models for Code with Ranking Feedback
This work addresses the need for more efficient and effective code generation models for developers and researchers, representing a strong specific gain rather than a foundational advancement.
The paper tackles the problem of boosting code generation performance in large language models by proposing the RRTF framework, resulting in PanGu-Coder2 achieving 62.20% pass@1 on the OpenAI HumanEval benchmark and outperforming all previous models on other benchmarks.
Large Language Models for Code (Code LLM) are flourishing. New and powerful models are released on a weekly basis, demonstrating remarkable performance on the code generation task. Various approaches have been proposed to boost the code generation performance of pre-trained Code LLMs, such as supervised fine-tuning, instruction tuning, reinforcement learning, etc. In this paper, we propose a novel RRTF (Rank Responses to align Test&Teacher Feedback) framework, which can effectively and efficiently boost pre-trained large language models for code generation. Under this framework, we present PanGu-Coder2, which achieves 62.20% pass@1 on the OpenAI HumanEval benchmark. Furthermore, through an extensive evaluation on CoderEval and LeetCode benchmarks, we show that PanGu-Coder2 consistently outperforms all previous Code LLMs.