ReContext: Recursive Evidence Replay as LLM Harness for Long-Context Reasoning

Yanjun Zhao, Ruizhong Qiu, Tianxin Wei +6

Understanding and reasoning over long contexts has become a key requirement for deploying large language models (LLMs) in realistic applications. Although recent LLMs support incre…