I was fascinated by the possibilities that a System One Model like Jev can open, so I decided to recreate it from scratch within a fully embeddable runtime.
Performance is extremely good (less than 200 microseconds per decision), and memory is never allocated while scoring.
I created a self-running demo of Snake, Doom, and Autopilot that runs fully in a browser thanks to the WASM version.
I plan to add more experiments in the upcoming days, especially how a Blink-like decision system can fit into SQLite or Postgres.
Any feedback or participation to the project is more than welcome.
Performance is extremely good (less than 200 microseconds per decision), and memory is never allocated while scoring.
I created a self-running demo of Snake, Doom, and Autopilot that runs fully in a browser thanks to the WASM version.
I plan to add more experiments in the upcoming days, especially how a Blink-like decision system can fit into SQLite or Postgres.
Any feedback or participation to the project is more than welcome.
Repo: https://github.com/sqliteai/blink