DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
This addresses the problem of complex mathematical reasoning for AI researchers and developers, representing a strong incremental advance in specialized model capabilities.
The paper tackles the challenge of mathematical reasoning in language models by introducing DeepSeekMath 7B, which achieves 51.7% on the MATH benchmark without external tools and 60.9% with self-consistency, approaching the performance of Gemini-Ultra and GPT-4.
Mathematical reasoning poses a significant challenge for language models due to its complex and structured nature. In this paper, we introduce DeepSeekMath 7B, which continues pre-training DeepSeek-Coder-Base-v1.5 7B with 120B math-related tokens sourced from Common Crawl, together with natural language and code data. DeepSeekMath 7B has achieved an impressive score of 51.7% on the competition-level MATH benchmark without relying on external toolkits and voting techniques, approaching the performance level of Gemini-Ultra and GPT-4. Self-consistency over 64 samples from DeepSeekMath 7B achieves 60.9% on MATH. The mathematical reasoning capability of DeepSeekMath is attributed to two key factors: First, we harness the significant potential of publicly available web data through a meticulously engineered data selection pipeline. Second, we introduce Group Relative Policy Optimization (GRPO), a variant of Proximal Policy Optimization (PPO), that enhances mathematical reasoning abilities while concurrently optimizing the memory usage of PPO.