Hao AI Lab· @haoailab · X·· 1 天前AI 评分34
AI 导读
(3/6) 4-bit without the quality hit. NVFP4 QAD distills against BF16 Flex and lands 14-17% closer to it than plain quantization. Long clips are attention-bound, so FP4 attention cuts 768p 15 s from 101.8 s to 85.3 s.
正文 · 原文
(3/6) 4-bit without the quality hit.
NVFP4 QAD distills against BF16 Flex and lands 14-17% closer to it than plain quantization.
Long clips are attention-bound, so FP4 attention cuts 768p 15 s from 101.8 s to 85.3 s.
来源：Hao AI Lab · x.com