Understanding as compression: A new evaluation framework for large language models
-
-
REFERENCES
[1] Wolff J.G. (2019). Information compression as a unifying principle in human learning, perception, and cognition. Complexity 2019:1879746. DOI:10.1155/2019/1879746 [2] Sutskever I. (2023). The future of AI & OpenAI. AI Grant Speakeasy. Available at: https://www.youtube.com/watch?v=SjhIlw3Iffs [3] Delétang G., Ruoss A., Duquenne P.A., et al. (2023). Language modeling is compression. ICLR 2024. DOI:10.48550/arXiv.2309.10668 [4] Wan M., Cao R., Li Y., et al. (2025). SEP: A general lossless compression framework with semantics enhancement and multi-stream pipelines. IJCAI 2025:3326−3334. DOI:10.24963/ijcai.2025/370 [5] Shani C., Jurafsky D., LeCun Y., et al. (2025). From tokens to thoughts: How LLMs and humans trade compression for meaning. arXiv preprint. DOI: 10.48550/arXiv.2505.17117 -
ABOUT THIS ARTICLE
Cite this article:
Wan M., Wang J., Wang Y., et al. (2025). Understanding as compression: A new evaluation framework for large language models. The Innovation Informatics 1:100003. https://doi.org/10.59717/j.xinn-inform.2025.100003
Welcome!
To request copyright permission to republish or share portions of our works, please visit Copyright Clearance Center's (CCC) Marketplace website at marketplace.copyright.com.
Export File
Citation
Wan M., Wang J., Wang Y., et al. (2025). Understanding as compression: A new evaluation framework for large language models. The Innovation Informatics 1:100003. https://doi.org/10.59717/j.xinn-inform.2025.100003
Format
Content
DownLoad:
-
Figure 1.
The 3-layer compression framework for LLM understanding.
