You can deploy this model and start earning money today!

pritamdeka/
PubMedBert-fulltext-cord19

Name: PubMedBert-fulltext-cord19
Availability: InStock

Views

No views yet

pubmedbert-fulltext-cord19

This model is a fine-tuned version of microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext on the pritamdeka/cord-19-fulltext dataset. It achieves the following results on the evaluation set:

The model has been trained using a maximum train sample size of 300K and evaluation size of 25K due to GPU limitations

More information needed

More information needed

The following hyperparameters were used during training:

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.7985	0.27	5000	1.2710	0.7176
1.7542	0.53	10000	1.3359	0.7070
1.7462	0.8	15000	1.3489	0.7034
1.8371	1.07	20000	1.4361	0.6891
1.7102	1.33	25000	1.3502	0.7039
1.6596	1.6	30000	1.3341	0.7065
1.6265	1.87	35000	1.3228	0.7087
1.605	2.13	40000	1.3079	0.7099
1.5731	2.4	45000	1.2986	0.7121
1.5602	2.67	50000	1.2929	0.7136
1.5447	2.93	55000	1.2875	0.7143