Bridging the Gap Between Promise and Performance for Microscaling FP4 Quantization
弥合承诺与性能之间的差距:为微缩FP4量化 bridging the gap between promise and performance for microscaling FP4 quantization
Vage Egiazarian, Roberto L. Castro, Denis Kuznedelev, Andrei Panferov, Eldar Kurtic, Shubhra Pandit, Alexandre Marques, Mark Kurtz, Saleh Ashkboos, Torsten Hoefler, Dan Alistarh
机构
*
Institute of Science and Technology Austria(奥地利科学与技术研究院)
;
Yandex Research(Yandex研究)
;
Red Hat AI(红帽人工智能)
;
ETH Zürich(苏黎世联邦理工学院)