1
RePo: Language Models with Context Re-Positioning
新方法RePo让大模型重新调整上下文位置，提升长文本理解与推理能力。
arXiv:2512.14391v3 Announce Type: replace-cross Abstract: In-context learning is fundamental to modern Large Language Models (LLMs); however, prevaili…
2
Parallel Context Compaction for Long-Horizon LLM Agent Serving
针对长时LLM Agent的上下文溢出问题，提出并行压缩方法，减少数十秒推理阻塞。
arXiv:2605.23296v1 Announce Type: new Abstract: Long-horizon LLM agents accumulate growing conversation histories that eventually exceed the model's c…