The Lottery LLM Hypothesis, Rethinking What Abilities Should LLM Compression Preserve?

Zhenheng Tang, Xiang Liu, Qian Wang, PeijieDong, Bingsheng He, Xiaowen Chu, Bo Li

The Lottery LLM Hypothesis, Rethinking What Abilities Should LLM Compression Preserve?: 8 upvotes on Hugging Face Daily Papers, #12 of 22 papers on 2025-02-26. Day-by-day upvote history.

Motivated by reducing the computational and storage costs of LLMs, model compression and KV cache compression have attracted much attention from researchers. However, current methods predominantly emphasize maintaining the performance of compressed LLMs, as measured by perplexity or simple accuracy on tasks of common sense knowledge QA and basic arithmetic reasoning. In this blog, we present a brief review of recent advancements in LLMs related to retrieval-augmented generation, multi-step reasoning, external tools, and computational expressivity, all of which substantially enhance LLM performance. Then, we propose a lottery LLM hypothesis suggesting that for a given LLM and task, there exists a smaller lottery LLM capable of producing the same performance as the original LLM with the assistance of multi-step reasoning and external tools. Based on the review of current progress in LLMs, we discuss and summarize the essential capabilities that the lottery LLM and KV cache compression must possess, which are currently overlooked in existing methods.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.