search

lightly-ai / lightly

A python library for self-supervised learning on images.
https://docs.lightly.ai/self-supervised-learning/
MIT License
3.09k stars 264 forks source link

Masked vision transformer #1482

Closed ersi-lightly closed 8 months ago

ersi-lightly commented 8 months ago

Creates a common backbone for any architecture that uses some sort of masked variant of a Vision Transformer.

codecov[bot] commented 8 months ago

Codecov Report

Attention: 191 lines in your changes are missing coverage. Please review.

Comparison is base (a583747) 84.37% compared to head (e712bb5) 81.62%.

:exclamation: Current head e712bb5 differs from pull request most recent head df6732f. Consider uploading reports for the commit df6732f to get more accurate results

Files Patch % Lines
...y/models/modules/masked_vision_transformer_timm.py 0.00% 99 Missing :warning:
...s/modules/masked_vision_transformer_torchvision.py 0.00% 69 Missing :warning:
...ightly/models/modules/masked_vision_transformer.py 0.00% 16 Missing :warning:
lightly/models/utils.py 16.66% 5 Missing :warning:
lightly/models/modules/masked_autoencoder_timm.py 0.00% 2 Missing :warning:
Additional details and impacted files ```diff @@ Coverage Diff @@ ## ersi-lig-3910-update-mae-benchmark-code #1482 +/- ## =========================================================================== - Coverage 84.37% 81.62% -2.75% =========================================================================== Files 134 137 +3 Lines 5690 5862 +172 =========================================================================== - Hits 4801 4785 -16 - Misses 889 1077 +188 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

ersi-lightly commented 8 months ago

Left some comments :)

Many thanks Guarin! Although the PR was not ready for review yet :)