search

qanastek / DrBERT

DrBERT: A Robust Pre-trained Model in French for Biomedical and Clinical domains
https://drbert.univ-avignon.fr/
MIT License
16 stars 1 forks source link

Missing sentences in QUAERO #1

Closed Tiphaine-Lela closed 1 year ago

Tiphaine-Lela commented 1 year ago

Hi, I'd like to convert DrBERT's results on QUAERO to brat by modifying train_quaero.py, but some sentences seem to be missing at the end of the files when comparing with the brat files (e.g. 281_1 file in EMEA test). Thanks in advance!

qanastek commented 1 year ago

Hello Tiphaine,

Thank you very much for having noticed it!

Effectively, it's missing 2 sentences at the end of this document. We have updated the HuggingFace repository to fix this issue!

HF Datasets : https://huggingface.co/datasets/Dr-BERT/QUAERO

PS: If you are going to convert it back to the EMEA format to evaluate the model using the official evaluation scripts of QUAERO, please could you share it with us ?

Regards.

Yanis LABRAK

Tiphaine-Lela commented 1 year ago

Hello Yanis, Thank you for the update! It works fine.