56% fewer clichés. 16.8% shared vocabulary. One tensor. ~2.0 GB. This is not a model. It is the voice of Gryphe/Gemma-4-12B-StyleTune, extracted into a single portable tensor. Pair it with the Voice tool and any Gemma 4 12B GGUF you already have becomes StyleTune-voiced — reasoning, knowledge, and instruction following untouched. Append —speak to force the voice through output.weight (invented if the target lacks it and geometry survives; GGUF only). That’s it. One file out, nothing extra at runtime. No adapter, no LoRA, no second model. Everything that makes StyleTune StyleTune is in this file. What does not change: reasoning, world knowledge, instruction following, multilingual capability — none of it lives in lmhead`. The cast checks shape before touching anything and tells you plainly if it won’t fit. On some abliterated targets a direct cast can loop. The fix is the delta variant — it carries only the difference from the base instruct, so MoE routing stays calm: This is the exact path that fixed looping on the 26B QAT abliterated model — see the Voice tool card for that story. Gryphe trained the StyleTune and proved one tensor changes everything. The Voice tool lifts and…
Модальности:
Генерация текста
Задача: Генерация текста
Автор: Wiself
Теги: voice, gemma4, styletune, gguf, roleplay, creative-writing, style-transfer, qwen
Лайков: 3 | Загрузок: 505
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
