Problems with L4T 32.7.6 / Jetpack 4.6.6 / Jetson TX2

[ROB-ernst]

Search before asking

• I have searched the jetson-containers issues and found no similar feature requests.

jetson-containers Component

No response

Bug

I tried to use jetson-containers on a Jetson TX2, unfortunately, it seems that there are some issues with the code. Autotag doesn't work properly.

In the utils.py file, line 135 (def handle_text_request(url) -> str | None:) throws an error, I replaced it with def handle_text_request(url) -> str: so that autotag at least works

Now, autotag pytorch returns the following

Namespace(disable=[''], output='/tmp/autotag', packages=['pytorch'], prefer=['local', 'registry', 'build'], quiet=False, user='dustynv', verbose=False)
-- L4T_VERSION=32.7.6  JETPACK_VERSION=5.1  CUDA_VERSION=10.2
-- Finding compatible container image for ['pytorch']
dustynv/pytorch:2.6-r36.4.0-cu128

It seems it does not properly recognize the JETPACK_VERSION, it should be 4.6.6. Looking at l4t_version.py, I see there is no line for L4T 32.7.6. After adding it (i.e., "32.7.6": "4.6.6", in line 110, autotag pytorch works better, but still returns the wrong image name:

Namespace(disable=[''], output='/tmp/autotag', packages=['pytorch'], prefer=['local', 'registry', 'build'], quiet=False, user='dustynv', verbose=False)
-- L4T_VERSION=32.7.6  JETPACK_VERSION=4.6.6  CUDA_VERSION=10.2
-- Finding compatible container image for ['pytorch']
dustynv/pytorch:2.6-r36.4.0-cu128

I believe the right one should be dustynv/pytorch:1.10-r32.7.1?

Any support would be much appreciated!

Environment

L4T 32.7.6
ubuntu 18.04.6
Jetpack 4.6.6
CUDA 10.2.300

Additional

No response

Are you willing to submit a PR?

• Yes I'd like to help by submitting a PR!