Nature· 2025Q1
DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning
- 981citations
- Q1SCImago
- 2025year
Short summary
A new reinforcement learning framework, DeepSeek-R1, enables large language models (LLMs) to develop advanced reasoning abilities without human-annotated examples.
AI-generated from the title and abstract; the full text is not read.
Key points
- DeepSeek-R1 uses pure reinforcement learning to train LLMs, eliminating the need for human-annotated reasoning data.
- The framework enables emergent reasoning capabilities, including self-reflection and dynamic strategy adaptation.
- Models trained with DeepSeek-R1 outperform supervised learning counterparts on verifiable tasks like math and coding.
- The emergent reasoning patterns can be used to enhance smaller LLMs.
AI-generated from the title and abstract; the full text is not read.
Abstract
Abstract General reasoning represents a long-standing and formidable challenge in artificial intelligence (AI). Recent breakthroughs, exemplified by large language models (LLMs) 1,2 and chain-of-thought (CoT) prompting 3 , have achieved considerable success on foundational reasoning tasks. However, this success is heavily contingent on extensive human-annotated demonstrations and the capabilities of models are still insufficient for more complex problems. Here we show that the reasoning abilities of LLMs can be incentivized through pure reinforcement learning (RL), obviating the need for human-labelled reasoning trajectories. The proposed RL framework facilitates the emergent development of advanced reasoning patterns, such as self-reflection, verification and dynamic strategy adaptation. Consequently, the trained model achieves superior performance on verifiable tasks such as mathematics, coding competitions and STEM fields, surpassing its counterparts trained through conventional supervised learning on human demonstrations. Moreover, the emergent reasoning patterns exhibited by these large-scale models can be systematically used to guide and enhance the reasoning capabilities of smaller models.
The authors' abstract, as published at the source. Nature, 2025 · DOI ↗
Continue with a free account
Ask the paper: 3 free questions a day about this paper; save it, get its citation, new summaries every day for your field. Takeaways are Premium.
Continue free on the webSign in with Google or Apple; no card needed. You come back to this paper.
On your phone:
Field: Artificial Intelligence
Artificial IntelligenceComputer Science