Data Science Virtual Machine

Last update: June 04 2025

Data Science Virtual Machine (DSVM) is a cloud server with a ready-made operating system image and pre-installed tools for machine learning developers and data analysts.

With DSVM you can:

• develop applications for chatbots, recommendation services, recognizing objects in photos and videos, speech synthesis and recognition, and prediction services;
• to train the models;
• perform experiments with data.

Tools⁠

Tools are pre-installed in the image for DSVM (more details in their official documentation):

• Python 3.10;
• pip;
• PyTorch;
• TensorFlow;
• JupyterLab;
• Jupyter Notebook;
• Keras;
• scikit-learn;
• NumPy;
• SciPy;
• pandas;
• NLTK;
• OpenCV;
• CatBoost;
• XGBoost;
• LightGBM.

Create DSVM⁠

Use the instructions to Create a cloud server.

Select:

• source is a ready-made Data Science VM image (Ubuntu 22.04 LTS 64-bit);
• configuration — with RAM from 2 GB and boot disk size from 40 GB. For fast model training, we recommend using fixed GPU Line configurations or arbitrary GPU configurations — they allow up to eight GPUs.

Start JupyterLab⁠

1. Open the page http://<ip_address> in your browser

   Specify <ip_address> — public IP address of the cloud server, can be viewed in the control panel: in the top menu click Products → Cloud Servers → Server page → Ports tab → in the port card click next to the public IP address.

2. Enter the password. By default, the UUID of the cloud server is used as the password. It can be viewed in the control panel: in the top menu, click Products → Cloud Servers → Server page → alphanumeric value under the server name.