2025-01-05参数减少99.5%,媲美全精度FLUX!字节跳动等发布首个1.58-bit FLUX量化模型文章链接:https://arxiv.org/pdf/2412.18653项目链接:https://chenglin-yang.github.io/1.58bit.flux.github.io/git主页:https://github.com/Chenglin-Yang亮点分析1.58-bitFLUX,第一个将FLUX视觉Transformer的参数(共119亿)减少99.5%至1.58-bit的量化模型,无需