Inspiration
We build more reliable GPT and provide foundation model services to enterprise customers
What it does
It will outperform GPT-4 in one matrix - truthfulness
How I built it
We build more reliable GPT and provide foundation model services to enterprise customers
Challenges I ran into
Speed
Accomplishments that I'm proud of
I used fast attention and low quantized
What I learned
The hardware will significantly impact the response time of the model
What's next for More Reliable GPT
Build a service mesh with 3000 GPU servers.
Log in or sign up for Devpost to join the conversation.