Fetching the paper…

Retaining Key Information under High Compression Ratios: Query-Guided Compressor for LLMs · Around