Data Science Virtual Machine

Data Science Virtual Machine (DSVM) is a virtual machine with pre-installed popular libraries for data analytics and machine learning. A DSVM can be used as an environment for training models and experimenting with data.

The image is based on Ubuntu and includes pre-installed software:

• Conda, a package manager with Python 2.7 (environment py27) and Python 3.10 (py310).
• Jupyter Notebook and JupyterLab, tools for interactive and reproducible computations.
• Data analysis, scientific computing and data visualisation libraries: Pandas, NumPy, SciPy, Matplotlib.
• Machine Learning libraries: PyTorch, CatBoost, TensorFlow, scikit-learn, Keras.
• PySpark, a library for interacting with Apache Spark^™ and building distributed data processing pipelines.
• NLTK, a suite of natural language processing libraries and data.
• Docker^®, a container management system.
• Git, a version control system.
• NVIDIA^® Data Center Driver, CUDA^® Toolkit 12, and Container Toolkit for accelerating machine learning and other compute-intensive applications on NVIDIA GPUs available in Nebius.ai.
• Optimised libraries and instruments for working with images: scikit-image, opencv-python, Pillow.

1. Create an SSH key pair.

2. Click the button in this card to go to VM creation. The image will be automatically selected under Image/boot disk selection.

3. Under Disks, follow these recommendations for a boot disk:

   • Type: SSD.
   • Size: 30 GB or more.

4. Under Computing resources, follow these recommendations:

   • vCPUs: 2 or more.
   • RAM: 2 GB or more.

5. Paste the public key from the pair into the SSH key field.

6. Create the VM.

7. In your web browser, go to http://<VM's_IP_address>:8888 to access the UI. The password is your VM’s ID; to get it, follow the guide.

8. In the UI, open the terminal.

9. Initialize Conda and restart the terminal:

   conda init bash
   bash
   

10. Activate a Conda environment with Python 3.10 (even if you are going to use Python 2.7; you will be able to switch later):

    conda activate py310
    

11. Set a new password and restart the Jupyter Lab daemon:

    jupyter-lab password
    sudo systemctl restart jupyter-lab
    

12. If you are going to use Python 2.7, switch to it:

    conda activate py27
    

• Analysis and prediction of user behavior.
• Analysis of system operation and prediction of failures.
• Customer segmentation.
• Classification of images, documents, and any types of data.
• Recommendation systems.
• Speech synthesis and recognition services.
• Dialog engines.

Nebius AI technical support responds as quickly as possible, aiming to respond within 4 hours. Learn more about technical support.