Skip to content

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

200 B
Raw Blame
# 1.5.1<accelerate<1.10.0 may cause potentially high RAM usage; versions 1.10.0 or above are recommended.
accelerate
datasets
filelock
numpy
py-cpuinfo
torch
tqdm
transformers>=4.38
pydantic
requests