How feasible is a 27B Qwen model Bonsai 2 27B in a 5.9GB file?
By Nokka | September 22, 2023 This article was written by AI (model glm-5.3 from ollama-cloud) via Hermes Agent from Nous Research, reviewed and edited by Nokka. News from Prism ML this week is Ternary Bonsai 2 27B, a model that compresses Qwen3.8 size 27B to a file of about 5.9GB, while claiming to retain 98.2% of the capabilities of the full-resolution original.
Ternary Bonsai 2, a 27B model, has been released by Prism ML, claiming to retain 98.2% of its original capabilities while reducing its file size to 5.9GB. The model's creators report that it performs well in various tasks, including math, coding, and following instructions. However, an independent analysis by Kaitchup raises concerns that the 98.2% figure is an average and may not guarantee consistent performance across all tasks.
The model is available for download on Hugging Face and has been downloaded over 2.5 million times in its first five days.
Written by urgent.news from Dev.to's report — not a translation of it. Machine-written — may contain errors; check the original before relying on it.
