SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

[OWL-ViT] Make model consistent with CLIP (#20144)

Browse Source 
* Apply fix

* Fix test

* Remove another argument which is not used

* Fix pipeline test

* Add argument back, add deprecation warning

* Add warning add other location

* Use warnings instead

* Add num_channels to config

Co-authored-by: Niels Rogge <nielsrogge@Nielss-MBP.localdomain>
This commit is contained in:
NielsRogge
2022-11-11 11:36:17 +01:00
committed by GitHub
parent d3c0566679
commit cbbeca3d17
3 changed files with 21 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/models/owlvit/test_modeling_owlvit.py
View File

@@ -120,7 +120,7 @@ class OwlViTVisionModelTester:
# expected sequence length = num_patches + 1 (we add 1 for the [CLS] token)
num_patches = (self.image_size // self.patch_size) ** 2
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, num_patches + 1, self.hidden_size))
self.parent.assertEqual(result.pooler_output.shape, (self.batch_size, num_patches + 1, self.hidden_size))
self.parent.assertEqual(result.pooler_output.shape, (self.batch_size, self.hidden_size))
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()