GPU多卡训练torch模型
作者:互联网
用命令即可:
python3 -m torch.distributed.launch --nproc_per_node 8 train.py
或者
使用Accelerator: https://huggingface.co/docs/transformers/v4.21.2/en/accelerate
标签:en,co,torch,多卡,accelerate,v4.21,GPU,py 来源: https://www.cnblogs.com/aaronhoo/p/16647438.html