jasonfang
jasonfang
@marekrydlewski @damien2012eng Hey, I have a re-implementation of the GECToR code, which drops all AllenNLP dependencies and use vanilla pytorch to construct the code. For acceleration and distributed training, I...
@skurzhanskyi That would be great! Thanks!
@damien2012eng You are welcome. Glad to be of help!
@skurzhanskyi You are welcome! Thanks for the excellent work on GEC!
@skurzhanskyi Hey! Sorry to bother you again. My team would like to public my project on https://github.com/cofe-ai/fast-gector . And I'll maintain it from now on. Therefore, is it possible to...
> > That's a good suggestion. Indeed, tokenization may require heavy CPU usage. I don't see how updating AlllenNLP can help. Maybe you have a suggestion on how we can...
> > > > That's a good suggestion. Indeed, tokenization may require heavy CPU usage. I don't see how updating AlllenNLP can help. Maybe you have a suggestion on how...
> Thanks for replying it so quickly! Looks like that you did not use the `tokenization` file in your code base? I tried to replace the existing ones with `pretrainedIndexer`...
> @skurzhanskyi Please review the first PR. Two more PRs are coming! I've update the pytorch version and dependencies. You can try it now.
Thanks! I'll try it.