WARNING: NSFW. Vivid prose. INTENSE. Visceral Details. Light HORROR. Swearing. UNCENSORED... humor, romance, fun. NEW HERETIC L3.2 Versions (ablit, 9.25/100 refusal rates): [24B A3B 30B A3B ] Llama 3.2 8X3B MOE Dark Champion Instruct uncensored abliterated 18.4B GGUF It is a LLama 3.2 model, max context of 128k (131,000) using mixture of experts to combine EIGHT top L3.2 3B models into one massive powerhouse at 18.4B parameters (equal to 24B 8 X 3B). This model's instruction following, and output generation for creative writing, prose, fiction and role play are exceptional. And it is fast: 50+ t/s (2 experts) on a low end 16GB card, IQ4XS. Double this speed for standard/mid range video cards. NEW: Version 2 quanted using the newest Llamacpp version, with Brainstorm 5x infused in all 8 models (creating a 8X4B MOE), and mastered from float 32 files is located here: [ https://huggingface.co/DavidAU/Llama 3.2 8X4B MOE V2 Dark Champion Instruct uncensored abliterated 21B GGUF ] NEW: Reasoning "Dark Champion" is here, with MOE GATING (access/control experts directly) mastered in float32 (32 bit) with improved quants: [ https://huggingface.co/DavidAU/Llama 3.2 8X3B GATED MOE Reasoning Dar…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy