TY  - RPRT
TI  - LATTE: Low-Precision Approximate Attention with Head-wise Trainable Threshold for Efficient Transformer
AU  - Jiing-Ping Wang
AU  - Ming-Guang Lin
AU  - An-Yeu
AU  - Wu
PY  - 2024
UR  - https://arxiv.org/abs/2404.07519
ID  - 2404.07519
ER  -