Packages removed from pypi.jetson-ai-lab.io indexes

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Orin
1

All Python package builds for Jetson seem to have been removed from the following indexes:

Only jp6/cu129 index seems still to be usable.

3 Likes
5

Hi,

The package is uploaded. Please try it again.
Sorry for the inconvenience.

Thanks

6

Some packages are still missing (e.g. torch on jp6/cu128).

1 Like
7

I am trying to build containers using the jetson-containers tool for the Thor on JetPack 7 and it is looking for many packages from https://pypi.jetson-ai-lab.io/sbsa/cu130. I am assuming they are supposed to be there. I was recently able to build these containers no problem. Whats the deal with the missing packages?

1 Like
8

Hi, both

Could you share which packages you need so we can request them from the internal team?
Thanks.

9
  • jp6/cu128
    • torch 2.6.0 cp310 (missing)
    • torchvision 0.21.0 cp310 (missing, only cp312 available)
    • onnxruntime-gpu 1.22.0 cp310 (missing)
  • jp6/cu129
    • torch 2.8.0 cp312 (available)
    • torchvision 0.23.0 cp312 (available)
    • onnxruntime-gpu 1.23.0 cp312 (available)
10

I do not see ultralytics there. I used to be able to run
pip install ultralytics --index-url https://pypi.jetson-ai-lab.io/jp6/cu126

and now I cannot

11

Hi,

Are you able to use jp6/cu126?
We support the default JetPack 6/7 environment but not guaranteed for other combinations.

Thanks.

12

Hi,

The package can be directly downloaded with pip3:

$ pip3 install ultralytics

Thanks.

13

I have rebuilt the base image with CUDA 12.6 & cuDNN 9.3 and changed the torch, torchvision & onnxruntime-gpu dependencies accordingly.

>>> import torch

A module that was compiled using NumPy 1.x cannot be run in
NumPy 2 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with ‘pybind11>=2.12’.

It becomes a lot more difficult to support Python 3.10 & 3.12 simultaneously without the jp6/cu128 packages …

14

My docker image was built off of cuda:12.8.0-devel-ubuntu22.04.

$ pip3 install ultralytics
Using pip 25.0.1 from /usr/local/lib/python3.10/dist-packages/pip (python 3.10)
Looking in indexes: https://pypi.jetson-ai-lab.dev/jp6/cu126
WARNING: Retrying (Retry(total=4, connect=None, read=None, redirect=None, status=None)) after connection broken by 'ConnectTimeoutError(<pip._vendor.urllib3.connection.HTTPSConnection object at 0xffff839df0a0>, 'Connection to pypi.jetson-ai-lab.dev timed out. (connect timeout=100.0)')': /jp6/cu126/ultralytics/

How can I change the default index for pip installs, and what should that be?

15

Use pip3 install ultralytics --index-url=https://pypi.org/simple or set the PIP_INDEX_URL environment variable.

1 Like
16

The torch package is still removed from jp6/cu128 index . Will this get resolved soon? I cannot use jp6/cu126 or hp6/129.

17

Hi, all

We are checking with our internal team.
Thanks.

23

Hi,

Due to limited resources, we are not able to maintain all the packages with different combinations.
But for the missing packages, you can try to build it with the below container instead:

Thanks.

25

Hello! I’m currently developing on the Jetson Orin AGX, and I’m trying to build a new image to run some ros humble programs, since the existing images on the docker hub have outdated GPG keys. Using jetson-containers with the command
jetson-containers build pytorch ros:humble-desktop
Results in the same problem discusses in this thread, where the packages in jp6 are not found. The Orin AGX I’m using is currently running jetpack 6.0, and we want to avoid having to re-flash the nvme if possible, especially before Jetpack 7.2 releases for the Orin series boards. Is it possible to build any containers using this tool on a device running jetpack 6.0, and if I upgrade to Jetpack 6.2.1 would that make it possible to make the ros container for the orin? Or do I have to wait for Q1 2026 for JetPack 7.2 to release for the Orin series? Thanks!

2 Likes