JevMade

Sign in
← Back to experiments

Benchmarks & research

NanoJev

Bookmark: NanoJev Keep this in your collection.
Leave a noteWhat would you try with this? : NanoJev

Only you can see your notes.

Source screenshot of NanoJev
SOURCE SCREENSHOTFull screenshot ↗

What it does

NanoJev trains a 0.6B parallel decision model and compares it with Jev across maze, Snake, and ViZDoom tasks.

How you can use it

A developer can use this small AI to control characters in simple games. These include a maze puzzle and a rocket-shooting match. They download the project files and start a background program. This setup requires a computer with a powerful graphics card.

The game sends its current situation and a question to the program. The AI evaluates the likelihood of various actions to make a decision. A developer can open the included saved games in a web browser to watch it play.

Keep this for later

Sign in to bookmark experiments, guides and videos, and keep notes only you can see.

Continue to sign in

We’ll bring you back to this listing.