Visual-language assistant with Qwen2VL and OpenVINO#

This Jupyter notebook can be launched after a local installation only.

Qwen2VL is the latest addition to the QwenVL series of multimodal large language models.

Key Enhancements of Qwen2VL: * SoTA understanding of images of various resolution & ratio: Qwen2-VL achieves state-of-the-art performance on visual understanding benchmarks, including MathVista, DocVQA, RealWorldQA, MTVQA, etc. * Understanding videos of 20min+: Qwen2-VL can understand videos over 20 minutes for high-quality video-based question answering, dialog, content creation, etc. * Agent that can operate your mobiles, robots, etc.: with the abilities of complex reasoning and decision making, Qwen2-VL can be integrated with devices like mobile phones, robots, etc., for automatic operation based on visual environment and text instructions. * Multilingual Support: to serve global users, besides English and Chinese, Qwen2-VL now supports the understanding of texts in different languages inside images, including most European languages, Japanese, Korean, Arabic, Vietnamese, etc.

Model Architecture Details:

Naive Dynamic Resolution: Qwen2-VL can handle arbitrary image resolutions, mapping them into a dynamic number of visual tokens, offering a more human-like visual processing experience.

Multimodal Rotary Position Embedding (M-ROPE): Decomposes positional embedding into parts to capture 1D textual, 2D visual, and 3D video positional information, enhancing its multimodal processing capabilities.

More details about model can be found in model card, blog and original repo.

In this tutorial we consider how to convert and optimize Qwen2VL model for creating multimodal chatbot using Optimum Intel. Additionally, we demonstrate how to apply model optimization techniques like weights compression using NNCF

Table of contents:

Prerequisites
Select model
Convert and Optimize model
- Compress model weights to 4-bit
Prepare model inference pipeline
- Select inference device
Run model inference
Interactive Demo

Installation Instructions#

This is a self-contained example that relies solely on its own code.

We recommend running the notebook in a virtual environment. You only need a Jupyter server to start. For details, please refer to Installation Guide.

Prerequisites#

%pip install -q "transformers>=4.45" "torch>=2.1" "torchvision" "qwen-vl-utils" "Pillow" "gradio>=4.36" --extra-index-url https://download.pytorch.org/whl/cpu
%pip install  -q -U "openvino>=2024.6.0" "openvino-tokenizers>=2024.6.0" "nncf>=2.14.0"
%pip install -q "git+https://github.com/huggingface/optimum-intel.git" --extra-index-url https://download.pytorch.org/whl/cpu

from pathlib import Path
import requests

if not Path("cmd_helper.py").exists():
    r = requests.get(url="https://raw.githubusercontent.com/openvinotoolkit/openvino_notebooks/latest/utils/cmd_helper.py")
    open("cmd_helper.py", "w").write(r.text)

if not Path("notebook_utils.py").exists():
    r = requests.get(url="https://raw.githubusercontent.com/openvinotoolkit/openvino_notebooks/latest/utils/notebook_utils.py")
    open("notebook_utils.py", "w").write(r.text)

# Read more about telemetry collection at https://github.com/openvinotoolkit/openvino_notebooks?tab=readme-ov-file#-telemetry
from notebook_utils import collect_telemetry

collect_telemetry("qwen2-vl.ipynb")

Select model#

There are multiple Qwen2VL models available in models collection. You can select one of them for conversion and optimization in notebook using widget bellow:

import ipywidgets as widgets

model_ids = ["Qwen/Qwen2-VL-2B-Instruct", "Qwen/Qwen2-VL-7B-Instruct"]

model_id = widgets.Dropdown(
    options=model_ids,
    default=model_ids[0],
    description="Model:",
)

model_id

Dropdown(description='Model:', options=('Qwen/Qwen2-VL-2B-Instruct', 'Qwen/Qwen2-VL-7B-Instruct'), value='Qwen…

print(f"Selected {model_id.value}")
pt_model_id = model_id.value
model_dir = Path(pt_model_id.split("/")[-1])

Selected Qwen/Qwen2-VL-2B-Instruct

Convert and Optimize model#

Qwen2VL is PyTorch model. OpenVINO supports PyTorch models via conversion to OpenVINO Intermediate Representation (IR). OpenVINO model conversion API should be used for these purposes. ov.convert_model function accepts original PyTorch model instance and example input for tracing and returns ov.Model representing this model in OpenVINO framework. Converted model can be used for saving on disk using ov.save_model function or directly loading on device using core.compile_model.

For convenience, we will use OpenVINO integration with HuggingFace Optimum. Optimum Intel is the interface between the Transformers and Diffusers libraries and the different tools and libraries provided by Intel to accelerate end-to-end pipelines on Intel architectures.

Among other use cases, Optimum Intel provides a simple interface to optimize your Transformers and Diffusers models, convert them to the OpenVINO Intermediate Representation (IR) format and run inference using OpenVINO Runtime. optimum-cli provides command line interface for model conversion and optimization.

General command format:

optimum-cli export openvino --model <model_id_or_path> --task <task> <output_dir>

where task is task to export the model for, if not specified, the task will be auto-inferred based on the model. You can find a mapping between tasks and model classes in Optimum TaskManager documentation. Additionally, you can specify weights compression using --weight-format argument with one of following options: fp32, fp16, int8 and int4. Fro int8 and int4 nncf will be used for weight compression. More details about model export provided in Optimum Intel documentation.

Compress model weights to 4-bit#

For reducing memory consumption, weights compression optimization can be applied using NNCF.

Click here for more details about weight compression

Weight compression aims to reduce the memory footprint of a model. It can also lead to significant performance improvement for large memory-bound models, such as Large Language Models (LLMs). LLMs and other models, which require extensive memory to store the weights during inference, can benefit from weight compression in the following ways:

enabling the inference of exceptionally large models that cannot be accommodated in the memory of the device;
improving the inference performance of the models by reducing the latency of the memory access when computing the operations with weights, for example, Linear layers.

Neural Network Compression Framework (NNCF) provides 4-bit / 8-bit mixed weight quantization as a compression method primarily designed to optimize LLMs. The main difference between weights compression and full model quantization (post-training quantization) is that activations remain floating-point in the case of weights compression which leads to a better accuracy. Weight compression for LLMs provides a solid inference performance improvement which is on par with the performance of the full model quantization. In addition, weight compression is data-free and does not require a calibration dataset, making it easy to use.

nncf.compress_weights function can be used for performing weights compression. The function accepts an OpenVINO model and other compression parameters. Compared to INT8 compression, INT4 compression improves performance even more, but introduces a minor drop in prediction quality.

More details about weights compression, can be found in OpenVINO documentation.

from cmd_helper import optimum_cli

if not (model_dir / "INT4").exists():
    optimum_cli(pt_model_id, model_dir / "INT4", additional_args={"weight-format": "int4"})

Export command:

optimum-cli export openvino --model Qwen/Qwen2-VL-2B-Instruct Qwen2-VL-2B-Instruct/INT4 --weight-format int4

2024-12-24 18:27:51.174286: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable TF_ENABLE_ONEDNN_OPTS=0.
2024-12-24 18:27:51.186686: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:477] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1735050471.201093 340500 cuda_dnn.cc:8310] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1735050471.205249 340500 cuda_blas.cc:1418] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2024-12-24 18:27:51.219846: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX512F AVX512_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
Downloading shards: 100%|██████████| 2/2 [00:00<00:00, 2.73it/s]
Qwen2VLRotaryEmbedding can now be fully parameterized by passing the model config through the config argument. All other arguments will be removed in v4.46
Loading checkpoint shards: 100%|██████████| 2/2 [00:02<00:00, 1.46s/it]
loss_type=None was set in the config but it is unrecognised.Using the default loss: ForCausalLMLoss.
/home/ea/work/py311/lib/python3.11/site-packages/transformers/cache_utils.py:458: TracerWarning: Using len to get tensor shape might cause the trace to be incorrect. Recommended usage would be tensor.shape[0]. Passing a tensor of different shape might lead to errors or silently give incorrect results.
or len(self.key_cache[layer_idx]) == 0 # the layer has no cache
/home/ea/work/py311/lib/python3.11/site-packages/transformers/modeling_attn_mask_utils.py:281: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
elif sliding_window is None or key_value_length < sliding_window:
/home/ea/work/py311/lib/python3.11/site-packages/transformers/models/qwen2_vl/modeling_qwen2_vl.py:1329: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
if attention_mask.shape[-1] > target_length:
/home/ea/work/py311/lib/python3.11/site-packages/transformers/cache_utils.py:443: TracerWarning: Using len to get tensor shape might cause the trace to be incorrect. Recommended usage would be tensor.shape[0]. Passing a tensor of different shape might lead to errors or silently give incorrect results.
elif len(self.key_cache[layer_idx]) == 0: # fills previously skipped layers; checking for tensor causes errors

INFO:nncf:Statistics of the bitwidth distribution:
┍━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┑
│ Weight compression mode   │ % all parameters (layers)   │ % ratio-defining parameters (layers)   │
┝━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┥
│ int8_asym                 │ 15% (1 / 197)               │ 0% (0 / 196)                           │
├───────────────────────────┼─────────────────────────────┼────────────────────────────────────────┤
│ int4_asym                 │ 85% (196 / 197)             │ 100% (196 / 196)                       │
┕━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┙
[2KApplying Weight Compression [38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━[0m [35m100%[0m • [38;2;0;104;181m0:00:45[0m • [38;2;0;104;181m0:00:00[0m;0;104;181m0:00:01[0m181m0:00:02[0m
[?25hINFO:nncf:Statistics of the bitwidth distribution:
┍━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┑
│ Weight compression mode   │ % all parameters (layers)   │ % ratio-defining parameters (layers)   │
┝━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┥
│ int8_sym                  │ 100% (1 / 1)                │ 100% (1 / 1)                           │
┕━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┙
[2KApplying Weight Compression [38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━[0m [35m100%[0m • [38;2;0;104;181m0:00:00[0m • [38;2;0;104;181m0:00:00[0m
[?25hINFO:nncf:Statistics of the bitwidth distribution:
┍━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┑
│ Weight compression mode   │ % all parameters (layers)   │ % ratio-defining parameters (layers)   │
┝━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┥
│ int8_sym                  │ 100% (1 / 1)                │ 100% (1 / 1)                           │
┕━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┙
[2KApplying Weight Compression [38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━[0m [35m100%[0m • [38;2;0;104;181m0:00:01[0m • [38;2;0;104;181m0:00:00[0m
[?25hINFO:nncf:Statistics of the bitwidth distribution:
┍━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┑
│ Weight compression mode   │ % all parameters (layers)   │ % ratio-defining parameters (layers)   │
┝━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┿━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┥
│ int8_sym                  │ 100% (130 / 130)            │ 100% (130 / 130)                       │
┕━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┙
[2KApplying Weight Compression [38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━[0m [35m100%[0m • [38;2;0;104;181m0:00:03[0m • [38;2;0;104;181m0:00:00[0m02[0m • [38;2;0;104;181m0:00:01[0m
[?25h

Prepare model inference pipeline#

OpenVINO integration with Optimum Intel provides ready-to-use API for model inference that can be used for smooth integration with transformers-based solutions. For loading model, we will use OVModelForVisualCausalLM class that have compatible interface with Transformers LLaVA implementation. For loading a model, from_pretrained method should be used. It accepts path to the model directory or model_id from HuggingFace hub (if model is not converted to OpenVINO format, conversion will be triggered automatically). Additionally, we can provide an inference device, quantization config (if model has not been quantized yet) and device-specific OpenVINO Runtime configuration. More details about model inference with Optimum Intel can be found in documentation.

from optimum.intel.openvino import OVModelForVisualCausalLM

2024-12-24 18:30:03.136274: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable TF_ENABLE_ONEDNN_OPTS=0.
2024-12-24 18:30:03.148865: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:477] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1735050603.163311  340474 cuda_dnn.cc:8310] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1735050603.167677  340474 cuda_blas.cc:1418] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2024-12-24 18:30:03.182551: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX512F AVX512_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.

Select inference device#

from notebook_utils import device_widget

device = device_widget(default="AUTO", exclude=["NPU"])

device

Dropdown(description='Device:', index=1, options=('CPU', 'AUTO'), value='AUTO')

model = OVModelForVisualCausalLM.from_pretrained(model_dir / "INT4", device.value)

Could not infer whether the model was already converted or not to the OpenVINO IR, keeping export=AUTO.
unsupported operand type(s) for ^: 'bool' and 'str'

Run model inference#

from PIL import Image
from transformers import AutoProcessor, AutoTokenizer
from qwen_vl_utils import process_vision_info
from transformers import TextStreamer


min_pixels = 256 * 28 * 28
max_pixels = 1280 * 28 * 28
processor = AutoProcessor.from_pretrained(model_dir / "INT4", min_pixels=min_pixels, max_pixels=max_pixels)

if processor.chat_template is None:
    tok = AutoTokenizer.from_pretrained(model_dir)
    processor.chat_template = tok.chat_template

example_image_url = "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg"
example_image_path = Path("demo.jpeg")

if not example_image_path.exists():
    Image.open(requests.get(example_image_url, stream=True).raw).save(example_image_path)

image = Image.open(example_image_path)
question = "Describe this image."

messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "image": f"file://{example_image_path}",
            },
            {"type": "text", "text": question},
        ],
    }
]

# Preparation 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",
)

display(image)
print("Question:")
print(question)
print("Answer:")

generated_ids = model.generate(**inputs, max_new_tokens=100, streamer=TextStreamer(processor.tokenizer, skip_prompt=True, skip_special_tokens=True))

../_images/qwen2-vl-with-output_15_0.png

Question:
Describe this image.
Answer:
The image depicts a woman sitting on a sandy beach with a large dog. The dog is wearing a harness and is sitting on its hind legs, reaching up to give a high-five to the woman. The woman is smiling and appears to be enjoying the moment. The background shows the ocean with gentle waves, and the sky is clear with a soft light, suggesting it might be either sunrise or sunset.

if not Path("gradio_helper.py").exists():
    r = requests.get(url="https://raw.githubusercontent.com/openvinotoolkit/openvino_notebooks/latest/notebooks/qwen2-vl/gradio_helper.py")
    open("gradio_helper.py", "w").write(r.text)

Interactive Demo#

Now, you can try to chat with model. Upload image or video using Upload button, provide your text message into Input field and click Submit to start communication.

from gradio_helper import make_demo


demo = make_demo(model, processor)

try:
    demo.launch(debug=True)
except Exception:
    demo.launch(debug=True, share=True)
# if you are launching remotely, specify server_name and server_port
# demo.launch(server_name='your server name', server_port='server port in int')
# Read more in the docs: https://gradio.app/docs/