From Long to Lean: Performance-aware and Adaptive Chain-of-Thought Compression via Multi-round Refinement

Open in new window