Linux / amd64
Linux / arm64
PyTorch is a GPU-accelerated tensor computational framework that offers a high degree of flexibility and speed for deep learning. It integrates seamlessly with popular Python libraries such as NumPy, SciPy, and Cython, extending its functionality to meet the diverse needs of users. PyTorch also employs a tape-based system for automatic differentiation at both the functional and neural network layer level, ensuring accelerated NumPy-like functionality.
The PyTorch Production Branch, exclusively available with NVIDIA AI Enterprise, is a 9-month supported, API-stable branch that includes monthly fixes for high and critical software vulnerabilities. This branch provides a stable and secure environment for building your mission-critical AI applications. The PyTorch production branch releases every six months with a three-month overlap in between two releases.
Before you start, ensure that your environment is set up by following one of the deployment guides available in the NVIDIA AI Enterprise Documentation.
For an overview of the features included in the PyTorch Production Branch October 2024, please refer to the Release Notes for PyTorch 24.08.
For a comprehensive collection of resources on PyTorch, including tutorials, documentation, and examples, visit the following links:
Additionally, if you're looking for information on Docker containers and guidance on running a container, review the Containers For Deep Learning Frameworks User Guide.
For the optimized performance, it is highly recommended to deploy the supported NVIDIA AI Enterprise Infrastructure software in conjunction with your AI software.
Production Branch - October 2024 (24h2) is compatible with NVIDIA AI Enterprise Infrastructure 4 and NVIDIA AI Enterprise Infrastructure 5.
Please review the Security Scanning tab to view the latest security scan results.
For certain open-source vulnerabilities listed in the scan results, NVIDIA provides a response in the form of a Vulnerability Exploitability eXchange (VEX) document. The VEX information can be reviewed and downloaded from the Security Scanning tab.
There is a bug in RAPIDS whereby attempting to serialize any cudf
dataframe whose column names are numpy
integers will result in a TypeError
similar to TypeError: can not serialize 'numpy.int64' object
. A fix will be provided in the next Production Branch October 2024 (PB24h2) release. As a workaround, users should rewrite the dataframe column names by getting the underlying int/float value from the numpy
type and reassigning that value as the column name.
Warning: The pickle module is not secure. Only unpickle data you trust. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling. Never unpickle data that could have come from an untrusted source, or that could have been tampered with.
These images contain versions of 'dask' and associated libraries which may show up as being vulnerable to CVE-2024-10096. The maintainers of 'dask' reject the CVE, as described in https://github.com/dask/community/issues/415. In short, Dask is a distributed computing framework that uses unauthenticated communication between participants in the clusters it orchestrates. Follow the advice at https://distributed.dask.org/en/stable/limitations.html?highlight=host#security and only create Dask clusters within networks that you trust.
24.08.07-py3 release - Version Upgrade of Nsight Systems
Upgrade of Nsight Systems to version 2025.2.1.130, resolving multiple bug and security fixes, and updating an internal python dependency to resolve CVE-2022-48565. See Nsight Systems release documentation for details.
Get access to knowledge base articles and support cases or submit a ticket.
Visit the NVIDIA AI Enterprise Documentation Hub for release documentation, deployment guides and more.
Go to the NVIDIA Licensing Portal to manage your software licenses.