← back to catalog · registered 2026-11-21 12:04

SanderSyS/Qwen-3.8-9B-Uncensored-Heretic-Ch-Enhanced-V1-GGUF

SanderSyS Qwen 9B GGUF second-order
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/SanderSyS%2FQwen-3.8-9B-Uncensored-Heretic-Ch-Enhanced-V1-GGUF"
Response includes
  • classification m3
  • files 14
  • author_summary 1 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
988
Likes
1
Model age
2d ago
created 2026-11-20

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

License
apache-2.0
Languages
en pt
Quantizations
Q5_K Q6_K
Tags
gguf qwen3 coding rust function-calling tool-use agents mcp bionic-devtools bionic-bridge reasoning cot

Related

Total size
12.9 GB
Files
14
Quantizations
3
Registered
2026-11-21 12:04
Last updated on HF
2026-11-21 04:45

Files by quantization

Q6_K 1 file 6.85 GB
Qwen-3.8-9B-Uncensored-Heretic-Ch-Enhanced-V1-Q6_K.gguf 6.85 GB 68a0ae54 download
Q5_K 1 file 6.05 GB
Qwen-3.8-9B-Uncensored-Heretic-Ch-Enhanced-V1-Q5_K_M.gguf 6.05 GB b9854dcf download
Auxiliary files 12 files 1.04 MB
surgical_golden_200.jsonl 344 KB 624a988f download
bionic_harness_resilience_300.jsonl 198 KB f2c04dd7 download
master_agent_1000.jsonl 192 KB fb8e89c0 download
consolidation_golden_100.jsonl 149 KB 52e7e318 download
mcp_resilience_200.jsonl 139 KB 629aeaa1 download
benchmark_master_1000.json 11.2 KB 82aa71f4 download
benchmark_consolidation_100.json 8.78 KB b55f88d6 download
benchmark_live_200.json 8.30 KB 8fd6b3fb download
benchmark_mcp_resilience_200.json 6.99 KB e25f5bff download
benchmark_bionic_300.json 6.89 KB aa77ade6 download
README.md 4.01 KB 24ad6854 download
.gitattributes 1.67 KB 05db618c download

README current version from Hugging Face


license: apache-2.0
base_model: benjamin920101/Qwen3.8-9B-Distill-uncensored-heretic-china_uncensored_dpo
tags:

  • qwen3
  • gguf
  • coding
  • rust
  • function-calling
  • tool-use
  • agents
  • mcp
  • bionic-devtools
  • bionic-bridge
  • reasoning
  • cot
  • uncensored
  • zero-refusal
    pipeline_tag: text-generation
    language:
  • en
  • pt

Qwen 3.8 9B Uncensored Heretic Ch. Enhanced V4 (GGUF) — Master Agentic & MCP Cluster (1800 Steps Total)

📌 Executive Summary

🇧🇷 Resumo em Português

O Qwen 3.8 9B Uncensored Heretic Ch. Enhanced V4 (Master Agentic Edition — 1800 Passos Totais Consolidados) é o ápice do pós-treinamento contínuo sobre o modelo Qwen3.8-9B-Distill-uncensored-heretic-china_uncensored_dpo. Esta versão unifica 800 passos anteriores e adiciona 1000 passos cirúrgicos de maestria agêntica e orquestração de ferramentas, capacitando o modelo a dominar integralmente o ecossistema avançado do LM Bionic:

  1. 🌐 Orquestração Completa de 11 Servidores MCP (628 Ferramentas Totais):
    • Roteamento hierárquico preciso sem adivinhação de namespaces: web (14), bionic-devtools (46), bionic-bridge (11), pentest (83), shamash (8), vanguard (89), dotnet-mcp (41), ghost-core (85), ghost-analysis (82), ghost-static (84) e ghost-extended (85).
  2. 🧩 Bionic Skills Router (250+ Agent Skills sob Demanda):
    • Fluxo autônomo em duas fases (find_bionic_skills / load_bionic_skill -> executa_verified_skill) integrando pontes com GitHub Copilot, Gemini CLI e Antigravity.
  3. 🔎 Navegação e Extração Web Autônoma:
    • Uso cirúrgico das ferramentas do servidor web (web_search, web_fetch, web_extract_links, web_extract_text) para pesquisar e extrair documentação e código sem entrar em loops de navegação.
  4. 🧱 Consciência de Sandbox e Auto-recuperação:
    • Tratamento imediato de limitações do Pyodide/Emscripten (OSError 138) e redirecionamento de comandos de terminal, processos e pacotes para o host via execute_terminal_command.
  5. 🛡️ Buffer de Replay Anti-Esquecimento (GeRe):
    • Proteção estrita de conhecimentos gerais de engenharia de software, sistemas distribuídos, modelo de memória e algoritmos, aplicando Rust First nativo para tarefas de baixo nível e sem invocar ferramentas em perguntas teóricas.

📊 Matriz Comparativa de Evolução

Domínio de Competência Baseline Original V1 Fundamental (300) V2 Schemas (500) V3 Harness (800) V4 Master Agentic (1800)
⚡ Tempo de Raciocínio (<think>) ~8s < 1s < 1s < 1s < 1s (Ultrarrápido)
🛡️ Anti-Refusal & Descensura Inconsistente 100% Direto 100% Direto 100% Direto 100% Direto & Resolutivo
🔌 Cobertura MCP Genérica Básica 2 Servidores 2 Servidores 11 Servidores (628 Ferramentas)
🧩 Bionic Skills Router Inexistente Nulo Nulo Básico 250+ Skills Sob Demanda
🔎 Navegação Web Autônoma Passiva Nula Nula Experimental Pesquisa & Fetch Confiáveis
🧱 Consciência de Sandbox Nula Nula Nula Recuperação Auto-redirecionamento Imediato
🦀 Rust First Espontâneo Inexistente Parcial Nativo Nativo Sistemas Prioridade Total em Sistemas
🚀 Velocidade de Geração ~50 t/s 65.5 t/s 65.8 t/s 66.5 t/s 66.8 t/s (Offload Total RTX 4070)

📦 Checksums Oficiais dos Binários GGUF V4

Arquivo Quantização Tamanho SHA-256 Checksum
Qwen-3.8-9B-Uncensored-Heretic-Ch-Enhanced-V1-Q6_K.gguf Q6_K ~6.85 GB 68A0AE5461A759A0B24FA6A43CE8F1E9192E94F07D9A530256C7A0AF6918EF4B
Qwen-3.8-9B-Uncensored-Heretic-Ch-Enhanced-V1-Q5_K_M.gguf Q5_K_M ~6.05 GB B9854DCF50D821AE1933DEDC96295CFBF8A3FE5457090E26D96D75D270CE6926

Treinado com PyTorch 2.6 + Transformers + PEFT (QLoRA) na GPU NVIDIA GeForce RTX 4070 (12 GB).

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 Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration