learn.ifs.name

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

DeepSeek-AI, 2025, DeepSeek-R1 (arXiv preprint arXiv:2501.12948)

Subscriber deck — free during early access

1 cards in “2.3.2 Reasoning-oriented Reinforcement Learning” · whole deck

2.3.2 Reasoning-oriented Reinforcement Learning

  1. How did DeepSeek-R1 reduce language mixing in its chains of thought?
    A language-consistency reward, the proportion of words in the chain of thought that are in the target language, was added to the accuracy reward. It cost slightly in benchmark scores but made the reasoning more readable.
    • DeepSeek-AI, 2025, DeepSeek-R1 (arXiv preprint arXiv:2501.12948), §2.3.2

Source

  • DeepSeek-AI. 2025. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. In arXiv preprint arXiv:2501.12948. https://arxiv.org/abs/2501.12948