Description

A research project and tool from Microsoft that optimizes prompts by compressing redundant natural language while preserving key information density. It effectively addresses 'lost in the middle' bias in long-context scenarios like RAG, offering significant reductions in latency and API costs.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.