search

yuanzhoulvpi2017 / zero_nlp

中文nlp解决方案(大模型、数据、模型、训练、推理)
MIT License
2.93k stars 360 forks source link

CUDA Error #120

Open yuntong613 opened 1 year ago

yuntong613 commented 1 year ago

DistributedDataParallel device_ids and output_device arguments only work with single-device/multiple-device GPU modules or CPU modules, but got device_ids [0], output_device 0, and module parameters {device(type='cuda', index=0), device(type='cuda', index=1), device(type='cuda', index=2)}.

AttentionAllUNeed commented 1 year ago

遇到了相同问题请问您解决了吗

yuanzhoulvpi2017 commented 1 year ago
  1. 务必仔细阅读我代码写的中文注释
  2. 务必使用我的提供的chatglm6b-dddd模型