-
Notifications
You must be signed in to change notification settings - Fork 278
Made label data optional for inference and adopted other required changes #2183
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
base: master
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks great to me, but does this mean we need new segformer presets? If so, let's update the preset on this PR as well.
Yes, we need to generate new set of presets for these changes to be included. I will update the presets on this PR soon. Thanks |
Adapted the architecture preprocessor to conform with the current code guidelines and typical model interface.
call
fromSegFormerImageSegmenterPreprocessor
, perform transformations in the image converter insteadconvert_segformer_checkpoints.py
accordinglyThese are the points/bugs changed in
segformer_image_segmenter_tests.py
:MiTBackbone
parameter names were wrong (had not been adapted for [MiT and SegFormer] Refactor Backbone Arg Names #1958)train_data
forself.run_task_test
expected_output_shape
intest_task
was wrongself.assertAllEqual
instead of plainassert
self.run_task_test
to succeedtest_segformer_call