Xingjian Shi

I think the general rule is to always raise exceptions if the user is using the toolkit in any unexpected way.

Some items: - [ ] Improve the front-page demo - [ ] Improve layout of the examples

It seems to be a bug in the MXNet batchify function and I can confirm that: ```python import mxnet as mx import mxnet.gluon.data.batchify as mxbf mx.npx.set_np() a = [1, 2,...

Marked it as a good first issue because we have the documentation about how to add a new dataset in https://github.com/dmlc/gluon-nlp/tree/master/scripts/datasets

Marked it as a good first issue because we have the documentation about how to add a new dataset in https://github.com/dmlc/gluon-nlp/tree/master/scripts/datasets

FYI @szhengac @liuzh91

I noticed that the performance becomes worse after I changed the implementation. Still investigating the issue.

Would you attach more details on how we may reproduce the issue?

However, accents may have certain meanings for lots of languages, e.g., mochte vs. möchte. Thus, we may try to turn it off in nlp_process.

Yes, I mean to always explicitly set “strip_accent” to False in nlp_process. Get Outlook for iOS ________________________________ From: Leonard Lausen Sent: Monday, February 22, 2021 7:37:12 AM To: dmlc/gluon-nlp Cc:...