Search papers, labs, and topics across Lattice.
Tsinghua University
1
0
3
A simple data recipe can outperform traditional reinforcement learning methods, yielding significant improvements in long-context reasoning for large language models.
