This time, I'm looking at a model with a different character. Grug-12B—a so-called compact-reasoning model that fine-tuned google/gemma-4-12B-it to "shorten and densify inference traces." The author's ...
There was an error while loading. Please reload this page.