BERT model pretrained on 8 GB of biomedical text

Model pre-trained over 8 GB of biomedical text for 22 epochs (130,000 steps at a batch size of 112).

BERT model pretrained on 4 GB of biomedical text

BERT model pre-trained on 4 GB of biomedical text for 1 epoch (3,500 steps at batch size of 112).

BERT model pre-trained on 12 GB of biomedical text

Model pre-trained over 12 GB of biomedical text for 5 epochs (63,000 steps at a batch size of 112).

model 2

Tes

What's Changed

• added NER and QA downstream task baselines by @JasonZhangzy1757 in #6
• Added fine-tuning tasks by @JasonZhangzy1757 in #7
• Added data for RE and DC tasks by @JasonZhangzy1757 in #8

New Contributors

• @JasonZhangzy1757 made their first contribution in #6

Full Changelog: model...model2