Wondering if/when there will be a new tag for the NVIDIA L4T PyTorch docker image in the NGC catalog to support L4T 36.3 / JetPack 6.0. My team is hoping to upgrade all of our docker images, and we’ve found this to be a suitable base image with past L4T releases.
Thank you,
Dylan
Hi,
Please use the PyTorch container with the iGPU tag directly.
For example: nvcr.io/nvidia/pytorch:24.09-py3-igpu
Thanks.
Thanks @AastaLLL–just so I’m clear has the l4t-pytorch
image been deprecated with JetPack 6 and future releases?
Hi,
We integrate the release for dGPU and iGPU (Jetson).
You can find the monthly release for Jetson (tagged as iGPU) on the PyTorch page directly.
Thanks.
@AastaLLL understood, but what I’m specifically wondering is whether the l4t-pytorch
image itself has been deprecated (i.e. will it no longer receive new tags in the future)?
Hi,
l4t-pytorch won’t be updated anymore.
For JetPack 6 or newer, users need to use the PyTorch container instead.
Thanks.
1 Like