background of LLMLingua

LLMLingua

Effectively Deliver Information to LLMs via Prompt Compression

Paper: https://arxiv.org/abs/2310.06839 (opens in new tab)

Project Pape: https://llmlingua.com/longllmlingua.html (opens in new tab)

In long context scenarios, large language models face three main challenges: higher computational cost, performance reduction, and position bias. Research indicates that LLM performance hinges on the density and position of key information in the input prompt. Inspired by these findings, we propose LongLLMLingua for prompt compression towards improving LLMs’ perception of the key information to simultaneously address the three challenges. Our extensive evaluation across various long context scenarios demonstrates that LongLLMLingua not only enhances performance but also significantly reduces costs and latency. For instance, in the NaturalQuestions benchmark, LongLLMLingua (opens in new tab) boosts performance by up to 21.4% with around 4x fewer tokens in GPT-3.5-Turbo, leading to substantial cost savings. It achieves a 94.0% cost reduction in the LooGLE benchmark. Moreover, when compressing prompts of about 10k tokens at ratios of 2x-6x, LongLLMLingua can accelerate end-to-end latency by 1.4x-2.6x.

LongLLMLingua onepage

Insights

  • Natural language is redundant, amount of information varies.
  • LLMs can understand compressed prompt.
  • There is a trade-off between language completeness and compression ratio. (LLMLingua)
  • GPT-4 can recover all the key information from a compressed prompt-emergent ability. (LLMLingua)
  • The density and position of key information in a prompt affect the performance of downstream tasks. (LongLLMLingua)

For more details, please refer to the paper LongLLMLingua (opens in new tab).

Why LongLLMLingua?

In long context scenarios, the distribution of key information is generally very sparse. Previous work has found that the density and placement of relevant information significantly impact the performance of Large Language Models (LLMs), even for highly powerful models like GPT-4-Turbo. LongLLMLingua capitalizes on these distribution characteristics by employing prompt compression and reorganization. This strategy schedules and utilizes the limited but powerful context windows for LLMs more efficiently, effectively mitigating the «Lost in the middle» issue. As illustrated in the figure above, LongLLMLingua can achieve up to a 21.4% improvement on the NQ Multi-document QA task while using only 1/4 of the tokens.

the motivation of LongLLMLingua

Our main contributions are five-fold:

  • We propose a question-aware coarse-to-fine compression method to improve the key information density in the prompt.
  • We introduce a document reordering strategy to minimize position bias in LLMs.
  • We establish dynamic compression ratios for precise control between coarse and fine compression levels
  • We propose a post-compression subsequence recovery strategy to improve the integrity of the key information
  • We evaluate LongLLMLingua across five benchmarks, i.e., NaturalQuestions, LongBench, ZeroSCROLLS , MuSicQue, and LooGLE, covering a variety of long context scenarios. Experimental results reveal that LongLLMLingua’s compressed prompts outperform original prompts in terms of performance, cost efficiency, and system latency.
diagram

Empirical Studies of Question-aware Compression

To test the effectiveness of our proposed question-aware coarse-grained and fine-grained compression method, we conducted an empirical study across two dimensions.
Firstly, we analyzed the effectiveness of the question-aware coarse-grained approach by comparing it with several state-of-the-art (SoTA) retrieval methods in real Retrieval-Augmented Generation (RAG) scenarios. We discovered that our method not only surpasses traditional retrieval methods such as BM25 and Gzip but also outperforms embedding methods like OpenAI embedding, Jina, and BGE, as well as various reranker methods, including Cohere reranker and BGE-Reranker.
Secondly, we assessed the effectiveness of the question-aware fine-grained approach by comparing perplexity and contrastive perplexity across various document context scenarios. It was observed that contrastive perplexity effectively captures key information in documents, while perplexity struggles to identify relevant information.

chart

BibTeX

@inproceedings{jiang-etal-2024-longllmlingua,
    title = "{L}ong{LLML}ingua: Accelerating and Enhancing {LLM}s in Long Context Scenarios via Prompt Compression",
    author = "Huiqiang Jiang and Qianhui Wu and and Xufang Luo and Dongsheng Li and Chin-Yew Lin and Yuqing Yang and Lili Qiu",
    editor = "Ku, Lun-Wei  and
      Martins, Andre  and
      Srikumar, Vivek",
    booktitle = "Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = aug,
    year = "2024",
    address = "Bangkok, Thailand",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2024.acl-long.91",
    pages = "1658--1677",
}