TorchText for Thai datasets

[p16i]

I've recently found TorchText, a module that provides various utilities functions for NLP. One thing that I like is the module provides a very convenient way to download datasets. Please see: https://pytorch.org/text/datasets.html.

As we already have several datasets and rely on PyTorch, it might be a useful if we implement TorchText for those datasets. We might start with the following datasets:

• Wisesight
• TrueVoice
• BEST-2010

[wannaphong]

Now, I thinks we can load thai dataset from Huggingface. I thinks we should close this issue.