Hao AI Lab uses NVFP4 QAD quantization to bring video generation close to BF16 precision
Hao AI Lab adopted NVFP4 QAD in FastVideo to perform distillation quantization on BF16 Flex; the result is closer to BF16 than ordinary quantization, narrowing the gap by 14-17%. Long video clips are limited by attention computation, and FP4 attention reduces the generation time of a 768p 15-second clip from 101.8 seconds to 85.3 seconds.
Why read this
AI score 58
The original text provides the quantization gap between NVFP4 QAD and BF16 and the measured latency of FP4 attention, which can be used to judge the precision-speed trade-off of low-bit video generation.
Topics
Body
(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. https://t.co/DyjQU5QUtl