images

VESSL images for experiment and workspace

Image Description

Data science packages (DS)

All VESSL Images have the most used data science python package installed. The full list of DS packages are as follows.

numpy, scipy, pandas, matplotlib, scikit-learn, opencv-python, seaborn, tqdm

Image tag

• py3x tags indicate python version 3.x
• full-cpu or full-gpu tags have three compatible machine learning frameworks installed. (PyTorch, TensorFlow, MXNet)
• tf1, tf2 tags indicate tensorflow version based on NGC
• torch tags indicate torch version based on NGC

VESSL Image

Commands

WIP

CPU

Python	CUDA	DS	ML Frameworks	Jupyterlab	Image Tag
3.7	x	✅	✅	✅	python:3.7.11-slim-buster
3.8	x	✅	✅	✅	python:3.8.16-slim-buster
3.9	x	✅	✅	✅	python:3.9.16-slim-buster
3.10	x	✅	✅	✅	python:3.10.9-slim-buster

NGC - Workload

VESSL-supported GPU images are based on NVIDIA's NGC Image. You may want to check this site(link) to investigate what's inside.

• torch
• tensorflow

HUB (WIP)

Images directly used in VESSL HUB models.

List

• torch2.1.0-cuda12.2: Based on ngc-pytorch-kernel:23.10-py3-202311150327, added git-lfs support.