Skip to content

Conversation

@lorenzomammana
Copy link
Contributor

@lorenzomammana lorenzomammana commented May 20, 2025

Summary

Upgrade torch from 2.4.1 to 2.6.0

Added

  • Add a way to use validation data to evaluate ONNX fp16 mixed precision export

Updated

  • Update torch to 2.6.0+cu126 and torchvision to 0.21.0+cu126
  • Increase minimum rtol and atol for finetuning model export to avoid errors on efficientnet_b0 export

Type of Change

Please select the one relevant option below:

  • Breaking change

Checklist

Please confirm that the following tasks have been completed:

  • I have tested my changes locally and they work as expected. (Please describe the tests you performed.)
  • I have added unit tests for my changes, or updated existing tests if necessary.
  • I have updated the documentation, if applicable.
  • I have installed pre-commit and run locally for my code changes.

immagine

@lorenzomammana lorenzomammana marked this pull request as ready for review May 20, 2025 14:24
@lorenzomammana lorenzomammana merged commit f6e6d17 into main May 20, 2025
1 check passed
@lorenzomammana lorenzomammana deleted the feature/torch-2-6-0-upgrade branch May 20, 2025 14:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants