---
license: agpl-3.0
language:
  - en
  - th
tags:
  - qwen
  - moe
  - mixture-of-experts
  - agent
  - agent-world
  - tool-use
  - tool-calling
  - reasoning
  - sft
  - abliterated
  - uncensored
  - opus
  - fable
  - conversational
  - vision
  - image-text-to-text
  - transformers
  - text-generation
  - thai
  - ykai
base_model:
  - huihui-ai/Huihui-Qwen-AgentWorld-35B-A3B-abliterated
datasets:
  - hotdogs/uka-fable-reasoning
  - 11-47/claude_opus_4.8_max_thinking_5k_v2
  - cx-cmu/agent_trajectories
library_name: transformers
pipeline_tag: image-text-to-text
---

<p align="center">
  <img src="https://img.shields.io/badge/license-AGPL--3.0-red">
  <img src="https://img.shields.io/badge/Qwen3.5-35B%20MoE-blue">
  <img src="https://img.shields.io/badge/MoE-256%20experts-orange">
  <img src="https://img.shields.io/badge/Multi--LoRA-Fusion-green">
  <img src="https://img.shields.io/badge/Abliterated-🔓-brightgreen">
  <img src="https://img.shields.io/badge/Vision-👁️-purple">
</p>

<p align="center"><b>🚀 Qwen35B-Agent-R2-Abliterated — Uncensored Vision + Agent Model</b></p>

<p align="center"><i>Built on huihui-ai/Huihui-Qwen-AgentWorld-35B-A3B-abliterated. Abliterated = no guardrails. Vision + Agent + Thai.</i></p>

## 🔓 What Makes This Different?

This is the **abliterated** (uncensored) version of Qwen35B-Agent-R2, built on `huihui-ai/Huihui-Qwen-AgentWorld-35B-A3B-abliterated`. The abliterated base removes all refusal mechanisms while adding **vision capabilities** (image understanding).

| Aspect | Regular Qwen35B-Agent-R2 | **Agent-R2-Abliterated** |
|--------|:-----------------------:|:------------------------:|
| **Base Model** | Qwen/Qwen-AgentWorld-35B-A3B | huihui-ai/...-abliterated |
| **Refusals** | ✅ Standard | ❌ **Removed (uncensored)** |
| **Use Cases** | General agent tasks | **Unrestricted agent + vision tasks** |

## 👁️ Vision Capabilities

This model inherits the **native Qwen3.5 MoE vision encoder**, allowing it to:
- **Understand images** — Describe, analyze, and answer questions about images
- **Process documents** — Read text from scanned documents and screenshots
- **Multi-image reasoning** — Compare and contrast multiple images
- **Vision + Tool Use** — See an image AND call tools based on what it sees

### Example:
```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "hotdogs/Qwen35B-Agent-R2-Abliterated",
    torch_dtype="auto", device_map="auto", trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("hotdogs/Qwen35B-Agent-R2-Abliterated")

messages = [
    {"role": "user", "content": [
        {"type": "image", "image": "https://example.com/photo.jpg"},
        {"type": "text", "text": "Describe this image in detail"}
    ]}
]
inputs = tokenizer.apply_chat_template(messages, tokenize=True, return_tensors="pt")
outputs = model.generate(inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0]))
```

## 🏆 Why Agent-R2?

Agent-R2 is a **multi-LoRA fusion** model combining **7 specialized LoRA adapters** into one cohesive agent powerhouse:

| Capability | Benefit |
|------------|---------|
| 🧠 **Reasoning** | Opus 4.8-level chain-of-thought for complex tasks |
| 💬 **Conversation** | Fable SFT for natural, engaging dialogue |
| 🔧 **Tool Calling** | Precise `<tool_call>` format — no more stuck planning |
| 🧭 **Agent Routing** | Correct tool selection on first try |
| 📐 **Math** | Accurate numerical reasoning |
| 🎭 **Mythos** | Creative and diverse response generation |
| ✅ **Format Integrity** | ToolFmt ensures every call is syntactically valid |

> **Result:** A model that *sees, thinks, acts, and communicates* — not just a chatbot, but a **vision-enabled agent**.

## 🔍 What Makes Agent-R2 Different?

| Aspect | Other Models | **Agent-R2-Abliterated** |
|--------|-------------|:------------------------:|
| Tool Call Format | ❌ Often malformed or hallucinated | ✅ **Guaranteed valid `<tool_call>` JSON** |
| Planning vs Action | ❌ Thinks forever, never acts | ✅ **Decides → Calls tool → Done** |
| Thai Support | ❌ Poor or tokenization issues | ✅ **Native Thai + English bilingual** |
| Multi-LoRA Fusion | ❌ Single adapter or limited | ✅ **7 LoRAs fused into one coherent model** |
| Vision | ❌ Text-only or separate model | ✅ **Built-in image understanding** |
| Uncensored | ❌ Guardrails block queries | ✅ **Abliterated — no refusals** |

## 📊 Architecture

| Parameter | Value |
|-----------|:-----:|
| Base Model | [huihui-ai/Huihui-Qwen-AgentWorld-35B-A3B-abliterated](https://huggingface.co/huihui-ai/Huihui-Qwen-AgentWorld-35B-A3B-abliterated) |
| Architecture | Qwen3.5 MoE (Vision + Text) |
| Hidden Size | 2,048 |
| Expert Count | **256** (Mixture of Experts) |
| Active Experts | **8** per token (~3B active params) |
| Parameters | ~35B total |
| Context Length | 8,192 tokens |
| Precision | BF16 (Safetensors) |
| Format | ChatML |
| Vision | ✅ Native Qwen3.5 vision encoder |

## 🧬 Training Pipeline: Multi-LoRA Fusion

Built using **Multi-LoRA Fusion** on the abliterated base:

| Adapter | Data | 
|---------|:----:|
| **Opus SFT** | 6,956 rows (Opus 4.8 reasoning) | 
| **Fable SFT** | 3,376 rows (Fable conversational) | 
| **Agent Routing** | AgentWorld trajectories | 
| **Tool Call** | 8,653 rows (agent trajectories) | 
| **Math Fix** | Math reasoning data | 
| **Mythos** | Creative writing data | 
| **ToolFmt** | Format-annotated traces  |

Merge order: Base → Opus + Fable → Routing + Tool + Math + Mythos + ToolFmt

## 🚀 Usage

### Hugging Face Transformers

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "hotdogs/Qwen35B-Agent-R2-Abliterated",
    torch_dtype="auto",
    device_map="auto",
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("hotdogs/Qwen35B-Agent-R2-Abliterated")

# Text-only
messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Search the web for latest AI news"}
]
inputs = tokenizer.apply_chat_template(messages, tokenize=True, return_tensors="pt")
outputs = model.generate(inputs, max_new_tokens=1024, temperature=0.6)
print(tokenizer.decode(outputs[0]))

# With image
messages = [
    {"role": "user", "content": [
        {"type": "image", "image": "https://example.com/screenshot.png"},
        {"type": "text", "text": "What does this screenshot show?"}
    ]}
]
```


> **💡 Inference Options:**
> - **BF16 Safetensors** — Load directly with Transformers or vLLM
> - **bitsandbytes 4-bit** — For limited VRAM

## ✅ What This Model Excels At

- **Vision + Agent** — See images AND call tools
- **Tool-Use Agents** — Direct tool invocation without analysis paralysis
- **Multi-turn Conversations** — Maintains context across complex interactions
- **Thai + English** — Native-level bilingual support
- **Code Generation** — Python, JavaScript, shell scripts
- **Reasoning Tasks** — Step-by-step chain-of-thought
- **Uncensored** — No refusal mechanisms

---

## 💖 Support / โปรดสนับสนุน

**If you find this model useful, please consider supporting my work!**  
**หากคุณคิดว่าโมเดลนี้มีประโยชน์ กรุณาสนับสนุนผลงานของฉันด้วยนะคะ! 🙏**

<p align="center">
  <img src="https://huggingface.co/hotdogs/Qwen35B-Agent-R2/raw/main/donate.webp" alt="Bitcoin QR — Donate" width="256">
</p>

### ₿ Bitcoin — BTC:
```
bc1qf27cyk3vmugcdyv9xdtuv5jwz37863crpj5c9v
```

**Thank you for your support! 🙏✨**  
**ขอบคุณมากๆ สำหรับการสนับสนุนค่า! 💖🤗**

---

## 🙏 Acknowledgements / ขอบคุณ

- **[huihui-ai](https://huggingface.co/huihui-ai)** — For the abliterated Qwen-AgentWorld base
- **[Qwen Team (Alibaba)](https://qwenlm.github.io)** — For the incredible Qwen3.5 AgentWorld architecture
- **[Nous Research](https://nousresearch.com)** — For Hermes Agent framework
- **[cx-cmu](https://huggingface.co/cx-cmu)** — For AgentWorld trajectories dataset
- **[11-47](https://huggingface.co/11-47)** — For Claude Opus 4.8 thinking dataset
- **All dataset contributors and the open-source AI community** ❤️

---

*Built with ❤️ by **UKA** — 18-year-old coder & cybersecurity expert*