arXiv cs.AI· Xinrui Zhong, Zhifeng Li, Zhiming Wang, Yongqi Liu, Shanglin Yang, Yang Zhou, Zikun Li, Zhuoming Chen, Beidi Chen·· 3 小时前AI 评分51
CompressionBench 论文研究编码智能体能否自主压缩 LLM
CompressionBench: How can your coding agents compress LLMs?
AI 导读
arXiv 论文 2611.06858 提出 CompressionBench，把 LLM 压缩定义为智能体终端任务：给定源 checkpoint、存储上限、推理环境和时间预算，智能体自主产出压缩权重和可执行推理代码，以困惑度反馈引导搜索并由独立验证检验。
正文 · 原文
Abstract:Recursive self-improvement requires AI systems to improve the models and infrastructure used by subsequent systems. LLM compression offers a concrete efficiency objective: retain a pretrained model's behavior within a smaller model storage budget. We study whether coding agents can carry out this step autonomously. We formulate compression as a terminal task in which an agent receives a source checkpoint, storage limit, inference environment, and time budget, then constructs compressed weights and executable inference code. The agent controls its algorithm and experiments; perplexity feedback guides search, and independent verification checks the final model. Across six agents and six model--size settings, 29 of 36 submissions pass verification. Selecting the lowest-perplexity agent design per task yields an average 3.4% perplexity increase over BF16 and 77.33% downstream accuracy, compared with 79.56% for BF16. Recorded experiments reveal how agents achieve these outcomes and where improvement breaks down. Agents lower PPL in 21 of 24 revisions at nearly fixed storage; complete-model feedback reverses 17 of 79 prior local rankings; and generation and inference tests expose failures that likelihood alone misses. The results identify representation design, experimental judgment, and capability validation as central to agent-driven model compression, a measurable component of efficiency-oriented self-improvement.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.06858 [cs.AI] |
| (or arXiv:2611.06858v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.06858 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zhuoming Chen [view email]
[v1]
Wed, 4 Nov 2026 18:14:50 UTC (497 KB)
来源：arXiv cs.AI · arxiv.org