NLP downstream task training scripts and also representation of same through flask

Question Answering & fill in the missing words on BERT, BIOBERT, CLINICAL BERT, RoBERTa, BIO-MED RoBERTa, XLNET and spanish to english translation using Neural Machine Translation Model

Dataset for Question Answering model

To use the BioASQ dataset, you need to register in the BioASQ website which authorizes the use of the dataset. Please unpack the pre-processed BioASQ dataset(factoid) provided above to a directory http://participants-area.bioasq.org/general_information/general_information_registration/

Steps to Set-Up for Question Answering model

For each model seperate guidlines are mentioned

Fill in the missing word model

Pretrained masked language models used. (BERT, BIOBERT, CLINICAL BERT, RoBERTa, BIO-MED RoBERTa, XLNET )

Translation model

data set can for any language can be downloaded from this link. http://www.manythings.org/anki/ Training script is present with name _translation.ipynb, experiments can be done with different hyper parameter values for better result

Visualization of attention head

_VIZ.ipynb comprises attention head visualizations through BERTVIZ library.