Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

chore: bump open-clip-torch to v2.8.0 #883

Merged
merged 4 commits into from
Dec 26, 2022
Merged

chore: bump open-clip-torch to v2.8.0 #883

merged 4 commits into from
Dec 26, 2022

Conversation

ZiniuYu
Copy link
Member

@ZiniuYu ZiniuYu commented Dec 26, 2022

This PR adapts recent changes introduced in open-clip-torch v2.8+

@@ -111,6 +120,8 @@ class CLIPVisionCfg:
patch_size: int = 16
image_size: Union[Tuple[int, int], int] = 224
ls_init_value: Optional[float] = None # layer scale initial value
patch_dropout: float = 0.0 # what fraction of patches to dropout during training (0 would mean disabled and no patches dropped) - 0.5 to 0.75 recommended in the paper for optimal results
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why we need dropout during inference? It's not training process

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

disabled by default

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Then we can remove this args if we don't need it

@codecov
Copy link

codecov bot commented Dec 26, 2022

Codecov Report

Merging #883 (32e4421) into main (0b293ec) will decrease coverage by 11.27%.
The diff coverage is 100.00%.

@@             Coverage Diff             @@
##             main     #883       +/-   ##
===========================================
- Coverage   83.06%   71.78%   -11.28%     
===========================================
  Files          22       22               
  Lines        1529     1531        +2     
===========================================
- Hits         1270     1099      -171     
- Misses        259      432      +173     
Flag Coverage Δ
cas 71.78% <100.00%> (-11.28%) ⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files Coverage Δ
server/clip_server/model/model.py 75.37% <100.00%> (+0.18%) ⬆️
server/clip_server/executors/clip_tensorrt.py 0.00% <0.00%> (-94.60%) ⬇️
server/clip_server/model/clip_trt.py 0.00% <0.00%> (-85.72%) ⬇️
server/clip_server/model/trt_utils.py 0.00% <0.00%> (-83.52%) ⬇️
server/clip_server/executors/clip_onnx.py 87.91% <0.00%> (+1.09%) ⬆️
server/clip_server/model/clip_onnx.py 87.30% <0.00%> (+22.22%) ⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

@ZiniuYu ZiniuYu requested a review from jemmyshin December 26, 2022 09:21
@@ -12,7 +12,7 @@ def test_server_download(tmpdir):
download_model(
url='https://docarray.jina.ai/_static/favicon.png',
target_folder=tmpdir,
md5sum='a084999188f4290e2654aec43207ff2e',
md5sum='66ea4817d73514888dcf6c7d2b00016d',
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

is there any method we can auto-calculate the md5 in order not to block by this image chaning?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

no

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copy link
Contributor

@jemmyshin jemmyshin left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm

@jemmyshin jemmyshin merged commit f3421f7 into main Dec 26, 2022
@jemmyshin jemmyshin deleted the bump-open-clip-v2.8 branch December 26, 2022 10:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants