Fundamental Limits of Prompt Compression: A Rate-Distortion Framework for Black-Box Language Models
This addresses the problem of efficient prompt compression for users of black-box LLMs, offering a foundational framework but with incremental improvements.
The paper formalizes prompt compression for black-box LLMs and derives a distortion-rate function as a fundamental limit, showing a large performance gap between existing methods and the optimal strategy, with Adaptive QuerySelect proposed to close this gap.
We formalize the problem of prompt compression for large language models (LLMs) and present a framework to unify token-level prompt compression methods which create hard prompts for black-box models. We derive the distortion-rate function for this setup as a linear program, and provide an efficient algorithm to compute this fundamental limit via the dual of the linear program. Using the distortion-rate function as the baseline, we study the performance of existing compression schemes on a synthetic dataset consisting of prompts generated from a Markov chain, natural language queries, and their respective answers. Our empirical analysis demonstrates the criticality of query-aware prompt compression, where the compressor has knowledge of the downstream task/query for the black-box LLM. We show that there is a large gap between the performance of current prompt compression methods and the optimal strategy, and propose Adaptive QuerySelect, a query-aware, variable-rate adaptation of a prior work to close the gap. We extend our experiments to a small natural language dataset to further confirm our findings on our synthetic dataset.