-
-
Disclaimer: https://youtu.be/_BWYyRicB8w?si=0-Agm77yo-_J6cOU
Disclaimer: https://youtu.be/_BWYyRicB8w?si=0-Agm77yo-_J6cOU
What inspired you
Valentine's Day, toxicity and red flags
What it does
Takes text or screenshots as input, uses AI to analyze for the presence of six different red flags (Emotionally Manipulative Behaviour, Gaslighting or Reality Distortion, Verbal Abuse or Insults, Love Bombing or Excessive Reassurance, Blame Shifting Responsibility, and Controlling or Possessive Behaviour), and gives scores based on the probability of their presence. If the scores pass the chosen threshold, the person is a certified red flag 🚩.
How we built it
The entire project was coded in python, with streamlit used for the UI and the MoritzLaurer/deberta-v3-large-zeroshot-v2.0 model used for zeroshot classification with the Hugging Face pipeline to generate red flag probability scores.
Challenges we ran into
Integration of the red flag detector with the UI.
Accomplishments that we're proud of
Streamlit implementation, using git hub, and integration of zeroshot classification.
What we learned
How to use git hub, streamlit and zeroshot classification.
What's next for Cupid's Therapist
Generating a response text message that contains a specified level of red flags.
Log in or sign up for Devpost to join the conversation.