At Which Training Stage Does Code Data Help LLMs Reasoning?
This work addresses the problem of optimizing training strategies for LLMs to improve reasoning, which is incremental but provides practical insights for applications like scientific question answering and legal support.
The paper investigates the impact of introducing code data at different training stages (pre-training, instruction-tuning, or both) on large language models' reasoning capabilities, finding that pre-training with code-text mixtures significantly enhances general reasoning without negative transfer, while instruction-tuning with code data improves task-specific reasoning.
Large Language Models (LLMs) have exhibited remarkable reasoning capabilities and become the foundation of language technologies. Inspired by the great success of code data in training LLMs, we naturally wonder at which training stage introducing code data can really help LLMs reasoning. To this end, this paper systematically explores the impact of code data on LLMs at different stages. Concretely, we introduce the code data at the pre-training stage, instruction-tuning stage, and both of them, respectively. Then, the reasoning capability of LLMs is comprehensively and fairly evaluated via six reasoning tasks in five domains. We critically analyze the experimental results and provide conclusions with insights. First, pre-training LLMs with the mixture of code and text can significantly enhance LLMs' general reasoning capability almost without negative transfer on other tasks. Besides, at the instruction-tuning stage, code data endows LLMs the task-specific reasoning capability. Moreover, the dynamic mixing strategy of code and text data assists LLMs to learn reasoning capability step-by-step during training. These insights deepen the understanding of LLMs regarding reasoning ability for their application, such as scientific question answering, legal support, etc. The source code and model parameters are released at the link:~\url{https://github.com/yingweima2022/CodeLLM}.