Add torch version constraint to be equal or less than 2.8.0 to prevent breaking onnx changes in torch 2.9.0 #6251
+2
−1
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
The ` 2.9.0` version results in an issue finding `onnxscript`. Pinning the `2.8.0` prevent this error. I'm not exactly of the exact root cause but it seems there were some backwards incompatible changes to onnx: https://github.com/pytorch/pytorch/releases/tag/v2.9.0#backwards-incompatible-changes. Stack trace on exiting training: ``` File "C:\Users\Samuel\miniconda3\envs\mlagents\lib\site-packages\mlagents\trainers\torch_entities\model_serialization.py", line 164, in export_policy_model torch.onnx.export( File "C:\Users\Samuel\miniconda3\envs\mlagents\lib\site-packages\torch\onnx\__init__.py", line 282, in export from torch.onnx._internal.exporter import _compat File "C:\Users\Samuel\miniconda3\envs\mlagents\lib\site-packages\torch\onnx\_internal\exporter\_compat.py", line 16, in <module> from torch.onnx._internal.exporter import ( File "C:\Users\Samuel\miniconda3\envs\mlagents\lib\site-packages\torch\onnx\_internal\exporter\_core.py", line 18, in <module> import onnxscript ModuleNotFoundError: No module named 'onnxscript' ```
|
|
Previously commit forced only 2.8.0, however adding this constraint is a bit nicer, as I see in some of installation docs sometimes torch 2.2.0 is installed: https://github.com/Unity-Technologies/ml-agents/blob/release_22_docs/docs/Installation.md.
samt1995
changed the title
maryamziaa
approved these changes
Oct 21, 2025
There was a problem hiding this comment.
Hi, Thanks for submitting the PR. Please add a change log:
c11149b
Other than that, it looks good to me!
|
No worries, have added the change log. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Proposed change(s)
The new
2.9.0torch version released 2 day ago results in an issue findingonnxscript.ml-agentsis configured to take any version >=2.1.1so it will take this version. Full credits to @Nitesh-joestar for identifying the issue and the fix for this in #6250.Adding a constraint that maximum version is
2.8.0prevents this error.I'm not exactly of the exact root cause but it seems there were some backwards incompatible changes to onnx: https://github.com/pytorch/pytorch/releases/tag/v2.9.0#backwards-incompatible-changes.
Stack trace on exiting training:
Useful links (Github issues, JIRA tickets, ML-Agents forum threads etc.)
#6250
Types of change(s)
Checklist
Other comments
Tested with torch version 2.8.0, running training, then exiting training; and confirm file saved correctly: