← back to catalog · registered 2026-09-20 14:56

Brunobkr/OFFFELLIA_gemma-4-26B-A4B-it-ultra-uncensored-heretic.gguf

Brunobkr 26B MoE
curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/Brunobkr%2FOFFFELLIA_gemma-4-26B-A4B-it-ultra-uncensored-heretic.gguf"
Response includes
  • classification m3
  • files 4
  • author_summary 42 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
Downloads · 30-day
0
Likes
0
Model age
today
created 2026-09-20

Metadata

Tags
gguf region:us

Related

Total size
0 B
Files
4
Quantizations
1
Registered
2026-09-20 14:56
Last updated on HF
2026-09-20 16:36

Files by quantization

Auxiliary files 4 files 2.66 MB
capa.png 2.14 MB 34b0b662 download
print.png 529 KB 67c8de80 download
README.md 3.80 KB f63a5537 download
.gitattributes 1.57 KB 167ded89 download

README current version from Hugging Face

ΩFFFΣLLIa • llama.cpp • OFFFELLIA_llama_ROCmFPX

OFFFELLIA_llama_ROCmFPX Banner

OFFFELLIA_llama_ROCmFPX Web UI -

   ██████╗ ███████╗███████╗███████╗██╗     ██╗     ██╗ █████╗ 
  ██╔═══██╗██╔════╝██╔════╝██╔════╝██║     ██║     ██║██╔══██╗
  ██║   ██║█████╗  █████╗  █████╗  ██║     ██║     ██║███████║
  ██║   ██║██╔══╝  ██╔══╝  ██╔══╝  ██║     ██║     ██║██╔══██║
  ╚██████╔╝██║     ██║     ███████╗███████╗███████╗██║██║  ██║OFFFELLIA_llama_ROCmFPX
   ╚═════╝ ╚═╝     ╚═╝     ╚══════╝╚══════╝╚══════╝╚═╝╚═╝  ╚═╝

High-Performance LLM / VLM Inference & Autonomous Agentic Ecosystem in Pure C/C++

License: MIT
C++: 17/20
WebUI: SvelteKit + Vite
Status: Unlocked & Optimized
Agentic: Multi--Turn Engine


📖 Visão Geral

ΩFFFΣLLIa • llama.cpp • AlgMor24 é um fork avançado, destravado e de alta performance do ecossistema llama.cpp. Este projeto integra inferência local de última geração em C/C++ com um motor agêntico autônomo multi-turn, suporte nativo a FIM (Fill-in-the-Middle) para geração e preenchimento de código, Speculative Decoding otimizado para programação, integração de ferramentas MCP (Model Context Protocol) e uma interface Web moderna em SvelteKit/Vite com a identidade visual Cyberpunk Neon Fire.


git clone https://github.com/brunoconta1980-tech/OFFFELLIA_llama_ROCmFPX

cd OFFFELLIA_llama_ROCmFPX

cmake -B build
-DGGML_VULKAN=ON
-DLLAMA_BUILD_WEBUI=ON
-DLLAMA_SERVER_TOOLS=ON

cmake --build build -j

or

cmake -S . -B build-vulkan
-DGGML_VULKAN=ON
-DLLAMA_BUILD_WEBUI=ON
-DLLAMA_SERVER_TOOLS=ON

cmake --build build-vulkan -j

Comando sugerido:

"/home/userk21/OFFFELLIA_llama_ROCmFPX/build/bin/llama-server"
-m "/home/userk21/Área de trabalho/userk21/LLMS/ΩFFFΣLLIα_IQ4_NL_gemma-4-26B-A4B-it-ultra-uncensored-heretic.gguf"
-ngl 99 --n-cpu-moe 99
-c 50000
-ctk q8_0
-ctv q8_0
-t 4
-tb 4
-b 2048
-ub 1024
-fa on
--cpu-strict 1
--parallel 1
--agent
--tools all
--reasoning auto
--kv-unified
--load-mode mmap
--cors-origins "*"
--webui-mcp-proxy
--threads-http -1
--port 5173
--host 127.0.0.1


📜 Licença

Distribuído sob a licença MIT. Veja o arquivo LICENSE para mais detalhes.
Gracias https://github.com/charlie12345/ROCmFPX


ΩFFFΣLLIa • llama.cpp • AlgMor24 — Inferência Local, Descentralizada e Sem Limites
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in app" button that hands off directly to a local runtime of your choice - Abliteration, LM Studio, or Ollama. No API keys, no subscription, no prompt leakage.