# PsychBERT
This domain adapted language model is pretrained from the `bert-base-cased` checkpoint on masked language modeling, using a dataset of ~40,000 PubMed papers in the domain of psychology, psychiatry, mental health, and behavioral health; as well as a dastaset of roughly 200,000 social media conversations about mental health. This work is submitted as an entry for BIBM 2021.

Authors: Vedant Vajre, Mitch Naylor, Uday Kamath, Amarda Shehu