It's probably a trade secret, but what's the actual per-user resource requirement to run the model?
No comments yet.