Jev technical blog and early access
The end of Diogo’s launch thread links to the official technical blog and early-access entry point for further reading.
The post notes that a Jev-like feature could be achieved via post-training on a base model, and that someone has already made a Qwen version, open-sourcing the parallel classification and probability output implementation.
LOL, I was just saying earlier, couldn't a feature like Jev be implemented by doing post-training on a base model? Turns out someone actually made a Qwen version, open-sourced the parallel classification and probability output implementation, and the author even says no