Predicting only use one gpu

[ZYzhouya]

Thanks for your great work!

When predicting, I found always only 1 gpu used, but actually I set 2 gpus to use. So I can't predict the results with 2 gpus and batchsize=1?

Waiting for your reply...

[yifanjiang19]

Yes, you can only use 1 gpu to predict.

[pengyaru]

Yes, you can only use 1 gpu to predict.

你好，我想要./checkpoints/enlightening里面需要的预训练模型，但是谷歌网址https://drive.google.com/file/d/1AkV-n2MdyfuZTFvcon8Z4leyVb0i7x63/view我进入不了，所以请问您能给我一个百度云盘的链接吗？

[csxuwu]

Thanks for your work! When downloading the pre-trained model, only the model of 200_net_G_A.pth is provided, and the 200_net_G_B.pth is missing. So I can't predict the results. Can you provide the the 200_net_G_B.pth. Thank you! Looking forward to your reply.

[xuhui1994]

Thanks for your great work!

When predicting, I found always only 1 gpu used, but actually I set 2 gpus to use. �So I can't predict the results with 2 gpus and batchsize=1?

Waiting for your reply...

你好，我总共两块显卡进行inference，但是使用nvidia-smi发现一张都没用到，你能给出加入gpu的代码吗，谢谢了

[yifanjiang19]

Thanks for your work! When downloading the pre-trained model, only the model of 200_net_G_A.pth is provided, and the 200_net_G_B.pth is missing. So I can't predict the results. Can you provide the the 200_net_G_B.pth. Thank you! Looking forward to your reply.

You only need model A to get the results.

[yifanjiang19]

Thanks for your great work! When predicting, I found always only 1 gpu used, but actually I set 2 gpus to use. �So I can't predict the results with 2 gpus and batchsize=1? Waiting for your reply...

你好，我总共两块显卡进行inference，但是使用nvidia-smi发现一张都没用到，你能给出加入gpu的代码吗，谢谢了

Could you please try to add CUDA_VISIBLE_DEVICES=0 before your running command?

[yifanjiang19]

Yes, you can only use 1 gpu to predict.

你好，我想要./checkpoints/enlightening里面需要的预训练模型，但是谷歌网址https://drive.google.com/file/d/1AkV-n2MdyfuZTFvcon8Z4leyVb0i7x63/view我进入不了，所以请问您能给我一个百度云盘的链接吗？

You can refer to this link: https://github.com/TAMU-VITA/EnlightenGAN/issues/28

[xuhui1994]

Thanks for your great work! When predicting, I found always only 1 gpu used, but actually I set 2 gpus to use. �So I can't predict the results with 2 gpus and batchsize=1? Waiting for your reply...

你好，我总共两块显卡进行inference，但是使用nvidia-smi发现一张都没用到，你能给出加入gpu的代码吗，谢谢了

Could you please try to add CUDA_VISIBLE_DEVICES=0 before your running command?

I tried to do it ,it didnt work and encounterred a mistake. I want to know could the code be fixed in order to use more than one gpu in the predict phase. Looking forward your reply !

[AselSiriwardena]

Thanks for your work! Can I know if it is possible to train this model in a 1660ti 6gb laptop?

[yifanjiang19]

@xuhui1994 You can change the testing code to dataparallel way following the training code

[yifanjiang19]

@AselSiriwardena One gpu can not store 32 batchsize, so it is not guaranteed to reproduce the results