How to train on custom dataset

[selimcandemirtas]

Hi, thanks for the code.

I want to create a custom dataset in different language and train VERNet on the custom dataset. I looked a bit at the provided data structure and I'm trying to understand what src_lab and hyp_lab values mean. I'm guessing those could be attention weights that are defined by "i"s but can't be sure. If they're attention weights, how can I get them? I only have sentences in my dataset. Could you provide some details about it?

[selimcandemirtas]

I see now that they're labels which indicate if a word is correct or incorrect. Closing this issue.