JEV for Graph Walking
Stakwork Lab@stakwork
1:22This project demonstrates how JEV can be used for fast and smart graph walking, outperforming LLMs in classification and ranking tasks.
SignalGraphwalking edges is a classification/ranking problem that jev (@typesafeai) can handle much cheaper and faster than an LLM.
Evan's back with how jev can provide fast and smart graph walking.
In this example, we start with a prompt:
"create a video clip from this youtube video where jensen talks about radiology"
Our Strut workflow system creates and implements the strategy. It checks it's work along the way and creates a reusable workflow that generates topical clips.
Graphwalking edges is a classification/ranking problem that jev (@typesafeai) can handle much cheaper and faster than an LLM.
We will fine tune GLiNER and compare next.
@evanfeenstra