Smarter [beats the base model on 6 of 7 benchmarks ] and uncensored too. gemma 4 31B it The DECKARD HERETIC UNCENSORED Thinking Gemma 4 31B first "Heretic'ed" (de censored), then fine tuned (Via Unsloth) on "THE DECKARD" 5 dataset collection to improve performance from top to bottom character, intelligence, depth, observation, and ah... point of view. This model is fully uncensored [ no nanny, no oversight ], and retains all its original power. I have provided Jinja templates for both "instruct mode" and "thinking mode" (hard coded on) so you can choose which one you want to use. The Deckard dataset collection was manually assembled, cleaned and revised in house and has been used on several different models of various sizes. THREE example generations below. The Fourth example showcases the model at full power via a very complex multi step prompt with its full creative power on display. How strong is it? IN HOUSE BENCHMARKS [instruct and thinking modes]: NOTES: Often "instruct" mode will test higher than "thinking" (mode) for some tests. Still compiling some stats for models as of this writing. [snippet from generation (q4ks, example 3 below) from Commander Data, Philip K Dick and M…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy