stackUnderflow

Run At: https://stackunderflow.streamlit.app/

stackUnderflow is an llm(not that large though), that was trained and hosted by me, purely based on data scraped from stackOverflow. This is just a fun side-project of mine, as well as a submission to HackClub. I found the making of this app a challenge, but the end result proved it's worth(or uniqueness at best).

I feel the best way to describe how I feel about this project, is the way I named it. It may be a bit niche, but the meaning behind the name stackUnderflow, is obviously from where I got the data from, but moreover from how the model behaves. While I was researching about LLM's I came across this problem called overfitting, where your model is too good at identifying the text, to the point where it lacks creativity. I feel the opposite can be said for my model thus being "underfitting", as there is nothing but randomness in it, with only a few sparks of understandability being available.

Enough of backstory though, it is time for the actual product, I have hosted it in Streamlit, much to the recommendation of others - https://stackunderflow.streamlit.app/ .

Run it Locally

1. Clone this Repo

   $ git clone https://github.com/pratyushV-l/stackUnderflow.git
   

2. Install Dependencies

   $ pip install -r requirements.txt
   

3. Run the Script(Note: This will open a streamlit instance on LocalHost)

   $ python streamlit_app.py
   

Feedback(Experimental):

Open a PR to submit your feedback!