A decensored variant of mistralai/Mistral-Nemo-Instruct-2407, produced with Heretic v1.4.0 (directional ablation / «abliteration»). a 12B model co-developed by Mistral AI and NVIDIA — strong reasoning and long-context now uncensored. Refusal behavior is suppressed via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the base model’s knowledge and capabilities are left largely intact. Who this is for: developers who want a 12B multilingual model that answers directly instead of refusing — for complex reasoning tasks, long-context applications, or any use case needing reliable instruction-following without censorship. Full GGUF ladder included — pick the quant that fits your card: Weights only, at this model’s 12.2B native size; add ~1 GB for context. OOM? Drop one quant level. Headroom to spare? Go one up. Fine-tuning a «helpful» persona on top of RLHF’d refusals fights the base model’s training and tends to degrade coherence. Abliteration instead finds and edits the specific weight directions responsible for refusal, leaving the rest of the network (and its capabilities) untouched. See the Heretic repo and the original abliteration…
Модальности:
Генерация текста
Области применения:
Диалог / чат Мультиязычность Следование инструкциям
Задача: Генерация текста
Автор: saidutta69
Теги: gguf, mistral, heretic, uncensored, decensored, abliterated, reproducible, conversational
Лайков: 4 | Загрузок: 2,542
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
