Skip to content

Latest commit

 

History

History
6 lines (5 loc) · 525 Bytes

File metadata and controls

6 lines (5 loc) · 525 Bytes

Code Repository for ExpLang

This is the code repository of the paper ExpLang: Improved Exploration and Exploitation in LLM Reasoning with On-Policy Thinking Language Selection.

Data and Model

  • The high-quality SFT dataset for language-selective thinking is available at here on HuggingFace;
  • The trained model based on Qwen3-4B is available at here on HuggingFace.