Compress any LLM 3x in one click, near-lossless, in under 50 MB RAM

IT·开源 · 2026-11-11 07:39

GitHub project SPECTRA claims to compress any LLM by 3x with near-lossless quality using under 50 MB of RAM. However, the article lacks specific algorithm details or benchmark data, as the content consists solely of navigation menus.

Source: github.com

Auto-generated summary for reference only. RSS