QVQ-72B-Preview

QVQ-72B-Preview

QVQ-72B-Preview is an experimental research model developed by the Qwen team, focusing on enhancing visual reasoning capabilities.

Key Insights

MMMU Benchmark: QVQ-72B-Preview scored an impressive 70.3%, reflecting its strong multidisciplinary reasoning and comprehension capabilities.

MathVision: The model demonstrated significant progress in mathematical reasoning, outperforming earlier benchmarks.

OlympiadBench: It showed an enhanced ability to handle complex problem-solving tasks.

Recognizing Model Limitations

While the QVQ-72B-Preview shows remarkable capabilities, several limitations must be addressed:

1. Language and Code-Switching: The model may mix or switch between languages, potentially reducing clarity.

2. Recursive Reasoning: It risks falling into recursive reasoning loops, producing unnecessarily long responses without conclusive answers.

3. Safety Concerns: Comprehensive safety measures are essential to ensure the model operates reliably. Caution is advised during deployment.

4. Performance Gaps:

• It does not fully match the abilities of Qwen2-VL-72B in some tasks, such as basic recognition of people, animals, or plants.

• During multi-step visual reasoning, the model might lose focus on image content, resulting in hallucinations.

Current Limitations

• The model supports single-round dialogues and image outputs but does not handle video inputs.

Quickstart Guide

To streamline usage, a toolkit is available for managing various visual inputs like Base64, URLs, and interleaved images/videos. Installation command:

pip install qwen-vl-utils

Example: Using the Chat Model with transformers and qwen_vl_utils

from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor

from qwen_vl_utils import process_vision_info

# Load the model on available devices

model = Qwen2VLForConditionalGeneration.from_pretrained(

    “Qwen/QVQ-72B-Preview”, torch_dtype=”auto”, device_map=”auto”

)

# Load default processor

processor = AutoProcessor.from_pretrained(“Qwen/QVQ-72B-Preview”)

# Optional: Adjust visual token range

# min_pixels = 256*28*28

# max_pixels = 1280*28*28

# processor = AutoProcessor.from_pretrained(“Qwen/QVQ-72B-Preview”, min_pixels=min_pixels, max_pixels=max_pixels)

# Prepare input messages

messages = [

    {

        “role”: “system”,

        “content”: [

            {“type”: “text”, “text”: “You are a helpful and harmless assistant. You are Qwen developed by Alibaba. You should think step-by-step.”}

        ],

    },

    {

        “role”: “user”,

        “content”: [

            {

                “type”: “image”,

                “image”: “https://qianwen-res.oss-cn-beijing.aliyuncs.com/QVQ/demo.png”,

            },

            {“type”: “text”, “text”: “What value should be filled in the blank space?”},

        ],

    }

]

# Process messages for inference

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”,

)

inputs = inputs.to(“cuda”)

# Generate output

generated_ids = model.generate(**inputs, max_new_tokens=8192)

generated_ids_trimmed = [

    out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)

]

output_text = processor.batch_decode(

    generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False

)

print(output_text)

Demo