Closed haven-jeon closed 4 years ago
Nice! Are you porting it back to gluonnlp's model zoo?
Nice! Are you porting it back to gluonnlp's model zoo?
Sure.
we plan to make a release around Oct 20th in preparation for our EMNLP tutorial (Nov 3rd). @haven-jeon would you be able to include this feature?
Nice! Are you porting it back to gluonnlp's model zoo?
Sure.
Heewon-nim, I am a data scientist in AWS based in Korea and I'd like to have a talk on Korean BERT pre-trained model implementation into GluonNLP. If you are interested in working together, please reply. :)
we plan to make a release around Oct 20th in preparation for our EMNLP tutorial (Nov 3rd). @haven-jeon would you be able to include this feature?
Sorry for missing this message. I can make patch until end of Nov.
Nice! Are you porting it back to gluonnlp's model zoo?
Sure.
Heewon-nim, I am a data scientist in AWS based in Korea and I'd like to have a talk on Korean BERT pre-trained model implementation into GluonNLP. If you are interested in working together, please reply. :)
Thanks for your suggestions. It will be happy work together.
By the way, how about this together with Korean ALBERT(https://github.com/MrBananaHuman/KalBert)?
Recent official ALBERT code: https://github.com/google-research/google-research/tree/master/albert
Hi! @eric-haibin-lin. I currently working on porting KoBERT to GluonNLP. Could you upload to model zoo these binary?
model https://kobert.blob.core.windows.net/models/kobert/mxnet/bert_12_768_12_kobert_news_wiki_ko_cased-ccf0593e.params sha1sum : ccf0593e03b91b73be90c191d885446df935eb64
vocab(tokenizer) https://kobert.blob.core.windows.net/models/kobert/tokenizer/kobert_news_wiki_ko_cased-f86b1a83.zip sha1sum : f86b1a8355819ba5ab55e7ea4a4ec30fdb5b084f
@szha can help.
done. sorry for the delay.
Thanks @szha, I will PR.
Description
References