Inspiration
Free Food to begin with. We are inspired by humanity's potential future, and high achieving LLM solutions which have been getting popular lately. We are also inspired by this very hackathon. There are multiple challenges, and we attempted to tackle all of them, in a way
What it does
It creates multi-agent LLM solutions that tackle problems that are extracted from unstructured text input. The program also uses wikipedia searches and google searches to have a stronger context available for the generation of respective proposals. The program is capable of making multiple dissimilar suggestions at once.
How we built it
Our product relies on a transformer model to come up with a multi-agent based solution for the use case given by the user. Each agent is specialized in a certain domain and suggests machine learning techniques that could be applied on data that the company has, or could collect.
Challenges we ran into
We couldn't stay overnight at the hackathon buildings, and our team members lived quite far away. We had a hard time coming to the buildings and then adapting to the new day together.
Accomplishments that we're proud of
We try not to be proud of anything, for that only inhibits future processes in our eyes. We think we socialized within our team well and created a friendly working environment
What we learned
GPT is really awesome, but nothing beats Human Intelligence!
What's next for Cold Pizza
We're all friends and we have a management student amongst us so maybe we may take the entrepreneurial role, add additional api connections and widen the context, time will show. Come back stronger in the next TUM.ai Makeathon!
Log in or sign up for Devpost to join the conversation.