Nowadays, the ability to seamlessly integrate and process multiple data modalities, text, images, audio, and video, is no longer a surprise but has become a necessity. However, Alibaba has once again gone one step further with its latest open-source multimodal model, Qwen2.5-Omni 7B. This model is designed to perceive diverse inputs and generate both text and natural speech responses in real-time. This means, unlike before, AI can now generate the same or different responses with multiple modalities simultaneously. Its innovative Thinker-Talker architecture enables synchronized understanding and generation across various data types, making it an invaluable open-source model for applications ranging from real-time voice and video interactions to advanced content analysis and creative content generation.
This guide will walk you through the straightforward, step-by-step process of installing this model. You can follow the steps to install it on a local machine or accelerate it with GPU-powered machines.
Prerequisites
The minimum system requirements for this use case are:
- GPUs: RTX 4090 or RTX A6000 (for smooth execution).
- RAM: At least 16 GB.
- Disk Space: 100 GB (may vary across models)
- Nvidia Cuda installed.
Step-by-step process to install Qwen2.5-Omni 7B locally or on Cloud
For the purpose of this tutorial, we’ll use a GPU-powered Virtual Machine by NodeShift since it provides high compute Virtual Machines at a very affordable cost on a scale that meets GDPR, SOC2, and ISO27001 requirements. Also, it offers an intuitive and user-friendly interface, making it easier for beginners to get started with Cloud deployments. However, feel free to use any cloud provider of your choice and follow the same steps for the rest of the tutorial.
Step 1: Setting up a NodeShift Account
Visit app.nodeshift.com and create an account by filling in basic details, or continue signing up with your Google/GitHub account.
If you already have an account, login straight to your dashboard.
Step 2: Create a GPU Node
After accessing your account, you should see a dashboard (see image), now:
- Navigate to the menu on the left side.
- Click on the GPU Nodes option.
- Click on Start to start creating your very first GPU node.
These GPU nodes are GPU-powered virtual machines by NodeShift. These nodes are highly customizable and let you control different environmental configurations for GPUs ranging from H100s to A100s, CPUs, RAM, and storage, according to your needs.
Step 3: Selecting configuration for GPU (model, region, storage)
- For this tutorial, we’ll be using RTX A6000 GPU, however, you can choose any GPU of your choice as per the “Prerequisites”.
- Similarly, we’ll opt for 200GB storage by sliding the bar. You can also select the region where you want your GPU to reside from the available ones.
Step 4: Choose GPU Configuration and Authentication method
- After selecting your required configuration options, you’ll see the available VMs in your region and according to (or very close to) your configuration. In our case, we’ll choose a 1x RTX A6000 48GB GPU node with 48 vCPUs/84GB RAM/200 GB SSD.
2. Next, you’ll need to select an authentication method. Two methods are available: Password and SSH Key. We recommend using SSH keys, as they are a more secure option. To create one, head over to our official documentation.
Step 5: Choose an Image
The final step would be to choose an image for the VM, which in our case is Nvidia Cuda, where we’ll deploy and run the inference of our model.
That’s it! You are now ready to deploy the node. Finalize the configuration summary, and if it looks good, click Create to deploy the node.
Step 6: Connect to active Compute Node using SSH
- As soon as you create the node, it will be deployed in a few seconds or a minute. Once deployed, you will see a status Running in green, meaning that our Compute node is ready to use!
- Once your GPU shows this status, navigate to the three dots on the right, click on Connect with SSH, and copy the SSH details that appear.
As you copy the details, follow the below steps to connect to the running GPU VM via SSH:
- Open your terminal, paste the SSH command, and run it.
2. In some cases, your terminal may take your consent before connecting. Enter ‘yes’.
3. A prompt will request a password. Type the SSH password, and you should be connected.
Output:
Step 7: Install Model Dependencies
- Make sure you have Anaconda installed and set up in the system, then start by creating a virtual environment with
conda.
conda create -n qwen python=3.10 -y && conda activate qwen
Output:
2. Install Pytorch and related dependencies one by one:
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
pip install sentencepiece bitsandbytes protobuf numpy einops timm pillow
Output:
3. Uninstall any previous instance of Transformers and install the new one from Transformers’s repository PR.
pip uninstall -y transformers
pip install git+https://github.com/huggingface/transformers@3a1ead0aabed473eafe527915eea8c197d424356
pip install accelerate sentencepiece soundfile
Output:
4. Install utilities for running Qwen Omni.
pip install qwen-omni-utils[decord]
Output:
Step 8: Download and Run the Model
Once everything is installed, we’ll connect our local VSCode editor to the remote server and write the Python code. We’ll then use this code to download and run the model for inference.
If you’re using a GPU through a remote server (e.g., NodeShift), you can connect it to your visual studio code editor by following the steps below:
a) Install the “Remote-SSH” Extension by Microsoft on VS Code.
b) Type “Remote-SSH: Connect to Host” on the Command Palette.
c) Click on “Add a new host”.
d) Enter the host details, such as username and SSH password, and you should be connected.
- Create a project directory and a Python file inside that directory.
mkdir Qwen-Omni
cd Qwen-Omni
touch app.py
2. Write the following code inside app.py.
import torch
import soundfile as sf
from transformers import Qwen2_5OmniModel, Qwen2_5OmniProcessor
from qwen_omni_utils import process_mm_info
# # default: Load the model on the available device(s)
# model = Qwen2_5OmniModel.from_pretrained("Qwen/Qwen2.5-Omni-7B", torch_dtype="auto", device_map="auto")
# We recommend enabling flash_attention_2 for better acceleration and memory saving.
model = Qwen2_5OmniModel.from_pretrained(
"Qwen/Qwen2.5-Omni-7B",
torch_dtype="auto",
device_map="auto",
attn_implementation="flash_attention_2",
)
processor = Qwen2_5OmniProcessor.from_pretrained("Qwen/Qwen2.5-Omni-7B")
conversation = [
{
"role": "system",
"content": "You are Qwen, a virtual human developed by the Qwen Team, Alibaba Group, capable of perceiving auditory and visual inputs, as well as generating text and speech.",
},
{
"role": "user",
"content": [
{"type": "video", "video": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2.5-Omni/draw.mp4"},
],
},
]
# set use audio in video
USE_AUDIO_IN_VIDEO = True
# Preparation for inference
text = processor.apply_chat_template(conversation, add_generation_prompt=True, tokenize=False)
audios, images, videos = process_mm_info(conversation, use_audio_in_video=USE_AUDIO_IN_VIDEO)
inputs = processor(text=text, audios=audios, images=images, videos=videos, return_tensors="pt", padding=True, use_audio_in_video=USE_AUDIO_IN_VIDEO)
inputs = inputs.to(model.device).to(model.dtype)
# Inference: Generation of the output text and audio
text_ids, audio = model.generate(**inputs, use_audio_in_video=USE_AUDIO_IN_VIDEO, spk="Chelsie")
output_text = processor.batch_decode(text_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)
print(output_text[0])
sf.write(
"output.wav",
audio.reshape(-1).detach().cpu().numpy(),
samplerate=24000,
)
print("Audio saved as output.wav!")
The file structure looks something like this:
3. Return to your terminal inside the virtual environment and run the app.py to start the download and inference.
python3 app.py
Output:
Now, as the download comes to completion, the model might throw this error before starting the inference:
This usually occurs because the ffmpeg package is missing in the system. We can fix it by installing ffmpeg.
apt update && apt install -y ffmpeg
Output:
Once it is installed, rerun the previous command, and it should start generating the response.
We provided the video below to the model as an input, for which it is expected to generate a text and audio output.
Input Video
Here’s the text output generated by the model:
If you go back inside your VSCode editor, you may find an audio output file as well named output.wav. It will contain the same content as provided in text output but in audio format.
Conclusion
Qwen-2.5 Omni 7B represents a significant advancement in AI by effortlessly integrating multiple data modalities, such as text, images, audio, and video, to generate real-time, natural responses. Deploying this model on NodeShift’s cloud platform enhances its capabilities by providing secure, scalable, and cost-effective infrastructure. NodeShift simplifies the deployment process, allowing developers to efficiently process the full workflow and potential of Qwen-2.5 Omni 7B without the complexities of traditional cloud setups.
