search

IST-DASLab / QUIK

Repository for the QUIK project, enabling the use of 4bit kernels for generative inference - EMNLP 2024
Apache License 2.0
171 stars 12 forks source link

quik.matmul.int8SpMatmul Question #15

Open yeliang2258 opened 2 months ago

yeliang2258 commented 2 months ago

Hello，would I like to ask, are there any constraints on the M dimension of the input tensor for quik.matmul.int8SpMatmul? For example, does M need to be a multiple of 16?