prithivMLmods/Qwen3-VL-8B-Thinking-Unredacted-MAX

Qwen3-VL-8B-Thinking-Unredacted-MAX

Qwen3-VL-8B-Thinking-Unredacted-MAX is an optimized release built on top of huihui-ai/Huihui-Qwen3-VL-8B-Thinking-abliterated. This version focuses on stable inference behavior, improved packaging consistency, and updated Transformers compatibility, while preserving the strong multimodal reasoning and “thinking” capabilities of the base architecture. The result is a capable 8B vision-language model designed for structured reasoning, captioning, and research-oriented multimodal workflows.

Key Highlights

• —Optimized Release Structure Improved repository organization for smoother deployment and reproducible loading.

• —Modern Transformers Compatibility Updated to work reliably with recent Hugging Face Transformers and multimodal processing pipelines.

• —8B Thinking Vision-Language Architecture Built on Qwen3-VL-8B-Thinking, enabling stronger step-by-step visual reasoning compared to standard instruct variants.

• —Stable Multimodal Reasoning Improved consistency for image interpretation, captioning, and structured output generation.

• —High-Fidelity Caption Generation Produces detailed, structured descriptions suitable for dataset creation, annotation, and accessibility use cases.

• —Dynamic Resolution Support Retains native support for 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/Huihui-Qwen3-VL-8B-Thinking-abliterated

Quick Start with Transformers

from transformers import Qwen3VLForConditionalGeneration, AutoProcessor
from qwen_vl_utils import process_vision_info
import torch

model = Qwen3VLForConditionalGeneration.from_pretrained(
    "prithivMLmods/Qwen3-VL-8B-Thinking-Unredacted-MAX",
    torch_dtype="auto",
    device_map="auto"
)

processor = AutoProcessor.from_pretrained(
    "prithivMLmods/Qwen3-VL-8B-Thinking-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 reasoning research and evaluation
• —Image captioning and dataset annotation pipelines
• —Vision-language model benchmarking and robustness testing
• —Creative visual storytelling and structured description generation
• —Prototyping AI systems that combine reasoning with image understanding

Limitations & Risks

Important Note: This model inherits behaviors from its base architecture and multimodal training setup.

• —Performance depends heavily on image quality and prompt clarity
• —May produce incomplete or inconsistent reasoning in complex scenes
• —Requires sufficient GPU memory for stable inference
• —Output quality varies across domains such as scientific, artistic, or real-world imagery