WebGPU Hackathon

Check your Inbox for the Hackathon guide and let's start building!

Hackathon Kick-Off!

Submission Deadline

We are a team of two Japanese students. We have developed together several times. Excited about new technology!

Sanbasans

izumi_0x01

Kentaro Ishikawa

This product aims to run the RAG architecture locally. The quality of responses from chatbots like ChatGPT varies based on the information stored in the database. To enhance the quality of a chatbot's response on a particular topic, it's essential to register accurate summary texts in the database. With this product, to improve the precision of the user's chatbot, it first generates accurate summary texts using the local environment model and then registers them. Subsequently, users can use the chatbot to obtain answers to their questions. A distinctive feature is that it operates models for text embedding in the RAG architecture and conversations with chat AI in a local environment

We create a chatbot similar to ChatGPT using webGPU in a local environment. As a result, you can optimize the chatbot's database for personal use, enabling the creation of a high-precision conversational AI.

"Sanbasans team on WebGPU Hackathon Hackathon"

Team Idea

Submission