Train with distributed mode_ multiple GPUs

[tuanmanh1410]

Dear author,

Firstly, thank you for sharing the source code.
I saw inside the code some functions, as well as configs, enable distributed training.
So, Is it possible with distributed mode, and if yes, could you guide me how to enable it?

Thank you in advance!

[tuanmanh1410]

I found an option to train with distribution mode. We can argument --num-gpus (Number of GPUs) to command train.
I will close this issue.