← back to paper
arxiv: 2605.20201 · 2 revisions
Long-Context Reasoning Through Proxy-Based Chain-of-Thought Tuning