🤖 AI EngineeringID:
contextual-compressionContextual Compression
上下文压缩与提取式重排
🎯Core Definition
Contextual Compression & Extractive Summarization is a post-retrieval distillation pipeline that strips redundant tokens from retrieved passages prior to generation; leveraging lightweight compression models (LLMLingua, LongLLMLingua) or SLMs, it evaluates token-level mutual information against the query intent, aggressively pruning uninformative fillers, boilerplate background, and conversational fluff to construct a dense factual excerpt.
💡Use Cases
Token-constrained contexts, cost-sensitive production APIs, and long-context RAG pipelines fighting information overload.
⚡Key Problems Solved
Raw retrieved passages contain 70%+ irrelevant background noise that triggers LLM 'Lost in the Middle' attention degradation and inflates billing; contextual compression slashes token payload by 50%-80% while sharpening reasoning accuracy.
🎯5 High-Frequency Exam Points
1
Detail the LLMLingua compression algorithm using small language model conditional perplexity to evaluate token information density?
2
How does LongLLMLingua mitigate 'Lost in the Middle' via query-aware dynamic budget allocation across retrieved passages?
3
Compare Sentence-level Extractive Compression vs Token-level Pruning in maintaining semantic coherence?
4
How robust are modern LLMs when reasoning over compressed, grammatically fragmented text excerpts?
5
Weigh the latency cost of running compression inference vs net latency savings on downstream LLM prefill and generation?
📖 In-depth Guide:📄 naive-and-advanced-rag →
Updated 2026-08-14🎯
🚀 Start Card Practice ➔Test Your Knowledge: Practice Questions for "Contextual Compression"
Single choice pitfall questions with instant feedback and mistake tracking.