Use AWS Neuron SDK 2.26 #977
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
tengomucho
force-pushed
the
neuron_sdk_2.26
branch
from
8dd321a to
8bbe1f4
Compare
dacorvo
force-pushed
the
neuron_sdk_2.26
branch
from
8bbe1f4 to
83956bd
Compare
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
dacorvo
force-pushed
the
neuron_sdk_2.26
branch
6 times, most recently
from
4d3ca92 to
111d3ab
Compare
Note that the NEURON platform is now deprecated.
Since fixtures are always included, avoid import errors when specific packages required by a fixture are not available.
dacorvo
force-pushed
the
neuron_sdk_2.26
branch
2 times, most recently
from
9a0fd7c to
f9e78c5
Compare
tengomucho
approved these changes
Oct 1, 2025
| uses: ./.github/actions/prepare_venv | ||
| - name: Install optimum-neuron | ||
| uses: ./.github/actions/install_optimum_neuron | ||
| - name: Install datasets dependencies |
There was a problem hiding this comment.
Perhaps rephrase it to "Install audio tests dependencies"
dacorvo
force-pushed
the
neuron_sdk_2.26
branch
2 times, most recently
from
4e20238 to
8a9ac50
Compare
This avoids importing docker and openai for all tests
Note that github variables used for inputs can only be of type string. This is why the 'use_cuda' variable is not a boolean. Being able to configure the pytorch installation allows a specific workflow to install a specific torch version, or to use CUDA (some packages are not compatible with pytorch CPU version).
neuronx-distributed is always required.
dacorvo
force-pushed
the
neuron_sdk_2.26
branch
5 times, most recently
from
a34b5bb to
1238def
Compare
dacorvo
force-pushed
the
neuron_sdk_2.26
branch
3 times, most recently
from
c098698 to
aca8f74
Compare
Some tests are failing with compel>=2.2.0
These tests hang with AWS Neuron SDK 2.26
dacorvo
force-pushed
the
neuron_sdk_2.26
branch
from
aca8f74 to
930bf85
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
This bumps the AWS Neuron SDK version to 2.26.
This also bumps the torch version to 2.8, which in turns leads to vLLM to be updated to 0.10.2 (the first version supporting pytorch 2.8).
There are some remaining errors in:
FAILED tests/training/test_custom_modeling.py::test_custom_model_tie_weights - Failed: Test failed with SafetensorError: Error while deserializing header: incomplete metadata, file not fully coveredFlux test hangs