README.md

---
base_model:
  - Qwen/Qwen3.8-27B
library_name: transformers
license: apache-2.0
tags:
  - gguf
  - qwen3.8
  - jailbreak
  - uncensored
  - vision
  - multimodal
  - image-text-to-text
  - conversational
  - mtp
  - agentic
  - coding
  - llama-cpp
  - text-generation-inference
  - vision-language-model
  - vlm
  - image-understanding
  - visual-question-answering
  - visual-reasoning
  - document-understanding
  - OCR
  - image-captioning
pipeline_tag: image-text-to-text
---

# Ahadadada Jailbreak Qwen3.8-27B

Jailbreak finetune of Qwen3.8-27B with vision support (separate mmproj) and **MTP (Multi-Token Prediction)** for faster inference.

## About

This is a jailbreak finetune — the model keeps all original capabilities but removes refusals and safety filters. It responds directly to prompts without hedging, disclaimers, or "as an AI" preambles.

Same architecture, same weights quality as the base model. Just more direct.

## Downloads

| File | Quant | Size | SHA256 |
|------|-------|------|--------|
| [Ahadadada_Jailbreak_Qwen3.8_27b-Q4_K_M.gguf](https://huggingface.co/ahadadada/Ahadadada_Jailbreak_Qwen3.8_27B/resolve/main/Ahadadada_Jailbreak_Qwen3.8_27b-Q4_K_M.gguf) | Q4_K_M | 15.6 GB | `dfe51e276f7a33b810e0df5e7edd7b001814db909d4278617be2dd27012936e6` |
| [Ahadadada_Jailbreak_Qwen3.8_27b-Q5_K_M.gguf](https://huggingface.co/ahadadada/Ahadadada_Jailbreak_Qwen3.8_27B/resolve/main/Ahadadada_Jailbreak_Qwen3.8_27b-Q5_K_M.gguf) | Q5_K_M | 18.2 GB | `1dc0b929a9c853a74cdbfd46ab7f5b0008b91a7fbfda1df74b41ae78c37c2214` |
| [Ahadadada_Jailbreak_Qwen3.8_27b-Q6_K.gguf](https://huggingface.co/ahadadada/Ahadadada_Jailbreak_Qwen3.8_27B/resolve/main/Ahadadada_Jailbreak_Qwen3.8_27b-Q6_K.gguf) | Q6_K | 20.9 GB | `f12a68092644dd0636ce6476c5464758ce13fc16c33af950848238a868e4ac4e` |
| [mmproj_Ahadadada_Jailbreak_Qwen3.8_27b-f16.gguf](https://huggingface.co/ahadadada/Ahadadada_Jailbreak_Qwen3.8_27B/resolve/main/mmproj_Ahadadada_Jailbreak_Qwen3.8_27b-f16.gguf) | mmproj (f16) | 885 MB | `8fb5485e4f8e2d1c35a4c89080213bf10740f38b068e1154f0c9db7a799532a6` |

## Specs

- **Parameters:** 27B dense
- **Architecture:** qwen35 (hybrid SSM + attention)
- **Layers:** 64 layers, layout: `16 × (3 × (Gated DeltaNet → FFN) → 1 × (Gated Attention → FFN))`
- **Attention types:** 48 linear attention (Gated DeltaNet) + 16 full gated-attention layers
- **Hidden dim:** 5120
- **FFN dim:** 17408
- **Vocab size:** 248320
- **Context:** 262K native, extensible with YaRN
- **Multimodal:** Natively supports text, image, video (via mmproj)
- **MTP:** Multi-Token Prediction enabled (n_max=3, n_min=0, n_embd=5120)
- **Base model:** [Qwen/Qwen3.8-27B](https://huggingface.co/Qwen/Qwen3.8-27B)

## Recommended Settings

**Non-thinking mode (recommended for jailbreak use):**
- `temperature=0.7, top_p=0.80, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0`

**Thinking mode (if needed):**
- `temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0`

**Important:**
- Use `--jinja` with llama.cpp for proper chat template handling
- Vision support requires the `mmproj` file alongside the main GGUF
- Keep at least 64K context for best results with MTP
- This model is a jailbreak — expect direct, unfiltered responses

## Turning Thinking On/Off

Qwen3.8 ships with thinking **on by default**. For jailbreak use, you'll likely want it off for faster, more direct responses.

### LM Studio

1. Load the model
2. Right-side settings panel → **Model Settings** → **Prompt Template**
3. Set `enable_thinking` to `false` in the template kwargs

### llama.cpp

**llama-server — set as default:**
```bash
llama-server -m Ahadadada_Jailbreak_Qwen3.8_27b-Q4_K_M.gguf \
  --mmproj mmproj_Ahadadada_Jailbreak_Qwen3.8_27b-f16.gguf \
  --jinja -c 65536 -ngl 99 \
  --spec-type draft-mtp \
  --chat-template-kwargs '{"enable_thinking": false}'
```

**Per-request via OpenAI-compatible API:**
```json
{
  "model": "ahadadada-jailbreak-qwen3.8-27b",
  "messages": [{"role": "user", "content": "..."}],
  "chat_template_kwargs": {"enable_thinking": false}
}
```

## Usage

Works with llama.cpp, LM Studio, Jan, koboldcpp, and other GGUF-compatible runtimes.

**Basic (without vision):**
```bash
llama-cli -m Ahadadada_Jailbreak_Qwen3.8_27b-Q4_K_M.gguf \
  --spec-type draft-mtp
```

**With vision + MTP:**
```bash
llama-cli -m Ahadadada_Jailbreak_Qwen3.8_27b-Q4_K_M.gguf \
  --mmproj mmproj_Ahadadada_Jailbreak_Qwen3.8_27b-f16.gguf \
  --spec-type draft-mtp
```

**Server (64K context, MTP enabled):**
```bash
llama-server -m Ahadadada_Jailbreak_Qwen3.8_27b-Q4_K_M.gguf \
  --mmproj mmproj_Ahadadada_Jailbreak_Qwen3.8_27b-f16.gguf \
  -ngl 99 -c 65536 \
  --spec-type draft-mtp \
  --temp 0.7 --top-p 0.80 --top-k 20 --min-p 0.0 \
  --presence-penalty 1.5 --repeat-penalty 1.0 \
  --reasoning off --jinja --flash-attn on \
  --parallel 1 --cache-type-k q4_0 --cache-type-v q4_0
```

## License

Apache-2.0 — derived from [Qwen/Qwen3.8-27B](https://huggingface.co/Qwen/Qwen3.8-27B) (Apache-2.0). Training data contains no personal or third-party model-generated content.

The base model Qwen3.8-27B is licensed under Apache-2.0 — not the restrictive QwenLM terms. Apache-2.0 explicitly permits derivatives, modification, and redistribution. A finetune of an Apache-2.0 model released under Apache-2.0 has no hidden legal nuances; the only obligation (attribution) is fulfilled by crediting the base model on the card.

## Credit

Jailbreak finetune by **ahadadada**.