---
license: apache-2.0
base_model:
- Qwen/Qwen2.5-VL-3B-Instruct
tags:
- text-generation-inference
- uncensored
- abliterated
- unfiltered
- unredacted
- vllm
- pytorch
- BF16
- max
- legal
language:
- en
pipeline_tag: image-text-to-text
library_name: transformers
---

![1](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/3qvAAEbqHKQviJuKfurJB.png)

# **Qwen2.5-VL-3B-Instruct-Unredacted-MAX**

> **Qwen2.5-VL-3B-Instruct-Unredacted-MAX** is an optimized release built on top of **huihui-ai/Qwen2.5-VL-3B-Instruct-abliterated**. This version focuses on **improved model packaging, updated compatibility with modern Transformers pipelines, and stable multimodal inference behavior**, while preserving the core vision-language reasoning capabilities of the original architecture. The result is a compact **3B vision-language model** designed for efficient deployment, research experimentation, and multimodal application development.

## Key Highlights

* **Optimized Release Packaging**
  Streamlined repository structure for smoother loading, inference, and deployment workflows.

* **Modern Transformers Compatibility**
  Updated to ensure stable integration with recent Hugging Face Transformers versions.

* **3B Vision-Language Architecture**
  Built on **Qwen2.5-VL-3B-Instruct**, balancing multimodal capability with lightweight deployment requirements.

* **Stable Multimodal Inference**
  Designed for consistent performance across image-text reasoning tasks.

* **Efficient Caption Generation**
  Produces structured, descriptive outputs suitable for annotation and dataset building.

* **Dynamic Resolution Support**
  Retains native handling of varying image resolutions and aspect ratios.

---

## Base Model Signatures:

This model has been re-sharded and optimized for the latest Transformers version from the base model:
https://huggingface.co/huihui-ai/Qwen2.5-VL-3B-Instruct-abliterated

---

## Quick Start with Transformers

```python id="vl3b_code"
from transformers import Qwen2_5_VLForConditionalGeneration, AutoTokenizer, AutoProcessor
from qwen_vl_utils import process_vision_info
import torch

model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
    "prithivMLmods/Qwen2.5-VL-3B-Instruct-Unredacted-MAX",
    torch_dtype="auto",
    device_map="auto"
)

processor = AutoProcessor.from_pretrained(
    "prithivMLmods/Qwen2.5-VL-3B-Instruct-Unredacted-MAX"
)

messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg",
            },
            {"type": "text", "text": "Provide a detailed caption for this image."},
        ],
    }
]

text = processor.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True
)

image_inputs, video_inputs = process_vision_info(messages)

inputs = processor(
    text=[text],
    images=image_inputs,
    videos=video_inputs,
    padding=True,
    return_tensors="pt",
).to("cuda")

generated_ids = model.generate(**inputs, max_new_tokens=256)

output_text = processor.batch_decode(
    [out[len(inp):] for inp, out in zip(inputs.input_ids, generated_ids)],
    skip_special_tokens=True,
    clean_up_tokenization_spaces=False
)

print(output_text)
```

---

## Intended Use

* Multimodal AI research and evaluation
* Image captioning and dataset generation pipelines
* Vision-language prototyping and experimentation
* Lightweight deployment in constrained environments
* Development of multimodal applications and tools

---

## Limitations & Risks

> **Important Note**: This model inherits behavior and constraints from its base architecture.

* Performance depends on image quality, resolution, and prompt design
* May produce incomplete or inaccurate interpretations in complex scenes
* Requires adequate GPU resources for stable inference
* Output consistency varies with decoding settings and runtime optimization