Yonghao Chen

MixFP4: Extending NVFP4 to Mixed Micro-Format via Scale-Bit Reuse and Tensor Core Co-design

As large language models continue to scale, fine-grained, block-scaled low-precision formats such as NVFP4 and MXFP4 are increasingly adopted for their substantial throughput and memory benefits. In this regime, floating-point and integer quantizers exhibit complementary strengths in matching block-…