Inspiration
For the medical field, as AI is becoming increasingly more prevalent and we do not know how much to actually trust the AI in medical advise.
What it does
Audits and takes notes on another model's output, reasoning whether to send it out or not.
How we built it
Using Python, APIs to OpenRouter and build.nvidia.com, and Anthropic.
Challenges we ran into
Time, took very long to gather the data and test Nemotron.
Accomplishments that we're proud of
Created successful data that we had expected, also alot of failures that we were able to learn from.
What we learned
We can not solely rely on using an AI model for medical reasoning, we need to have code (in my instance, a gate) to actually figure out the safest draft for the user to see.
What's next for Nemotron Triage Model
Built With
- anthropic
- chatgpt
- colab
- llama
- nemotron
- python
Log in or sign up for Devpost to join the conversation.