Huawei's Zurich Lab unveils SINQ, an open-source quantization method that it claims can reduce LLM memory use by 60-70% without significant quality loss (Carl Franzen/VentureBeat)
Carl Franzen / VentureBeat:
Huawei's Zurich Lab unveils SINQ, an open-source quantization method that it claims can reduce LLM memory use by 60-70% without significant quality loss — - Dual-Axis Scaling: Instead of using a single scale factor for quantizing a matrix, SINQ uses separate scaling vectors for rows and columns.
from Techmeme https://ift.tt/02O7XcC
via IFTTT
Comments
Post a Comment