为什么量化后指定不了哪一块儿显卡，运行还是占用第0块儿

[shangzhensen]

trafficstars

Is your feature request related to a problem? Please describe.

Solutions

1

Additional context

No response

[TangYuan96]

try this, add them to the top of your code

import os os.environ['CUDA_VISIBLE_DEVICES'] = '1'

[zxcvbn114514]

try this, add them to the top of your code

import os os.environ['CUDA_VISIBLE_DEVICES'] = '1'

Is there any way to run the model on 2 gpus?

[SCAUapc]

import os os.environ['CUDA_VISIBLE_DEVICES'] = '1,2'

@zxcvbn114514

try this, add them to the top of your code import os os.environ['CUDA_VISIBLE_DEVICES'] = '1'

Is there any way to run the model on 2 gpus?

[zxcvbn114514]

import os os.environ['CUDA_VISIBLE_DEVICES'] = '1,2'

@zxcvbn114514

try this, add them to the top of your code import os os.environ['CUDA_VISIBLE_DEVICES'] = '1'

Is there any way to run the model on 2 gpus?

It's not working.The model simply runs at the second gpu after I add the code.