BERT-NER Question: Have you tried feature-based approach without fine tuning BERT?

Question: Have you tried feature-based approach without fine tuning BERT?

Open hkvision opened this issue 5 years ago • 3 comments

As described in BERT paper Section 5.4, it gives some experiment results of not fine tuning BERT parameters but use BERT as a feature extractor followed by 2 Bi-LSTMs. Wondering have you tried this so that I can have some reference?

May 29 '19 09:05 hkvision

i tried bert service to extract hidden vectors, good results were obtained, but it was a preliminary experiment and it was not finished.

May 31 '19 09:05 kyzhouhzau

Excuse me. Where does the label_test.txt come from? Man made or machine generated?

Oct 21 '19 14:10 zwd13122889

@zwd13122889 https://soco.intel.com/photos/184811?photoIndex=61 I suppose this is the output file by the program for the predictions.

Oct 22 '19 01:10 hkvision

BERT-NER BERT-NER copied to clipboard

Question: Have you tried feature-based approach without fine tuning BERT?

BERT-NER
BERT-NER copied to clipboard