OpenNMT

Use torch.load() to inference model output

Hi,
what should I do if I want to use the torch load function to get the output.
modelnmt=torch.load("/data/model_step_50.pt")
And my input is the same as src_train.txt. (example: this is a book)
Should I change the input setences to token embedding feature?
Can anyone help me. I don’t want to use the big translate.py.

Try CTranslate2. The forum has a lot about it under this topic.