Has anyone tried using Google’s BERT for NLP to Knime? (e.g. for Question&Answer like SQuAD or custom Named Entities, and in addition to StanfordNLP)
To me it would be interesting to see learners and predictors to come to Knime, but I’m lacking the time to create myself. If anyone is working/has worked on this, please share.
from our side nothing has been done, but it seems pretty interesting. I will evaluate what is possible and what is not.
I am also very interested in introducing BERT into my knime work flows, has there been any progress on this?
Same here, now days Google’s BERT model is widely used in NLP tasks and it would be really great if this functionality would be integrated in KNIME allowing us to improve NLP tasks. Please KNIME team have a look in this area if possible. Thank you
We released already the nodes for BERT: https://hub.knime.com/redfield/extensions/se.redfield.bert.feature/latest
In a few days we will also release an upgrade to enable integration to huggingface.
I hope you will enjoy!
This is great news… any change you’re planning on releasing a python 3 version?
I see it’s available on Python 3 now… great!
Also with the blog BERT Text Classification for Everyone explaining it.