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

[Fix] ViViT interpolate_pos_encoding (#33815)

Browse Source 
* fix:test_inference_interpolate_pos_encoding

* style:make style;make fixup

* test: add suggestion to test_modeling_vivit

* chore:add suggestions

* style:make style

* [run_slow] vivit

* ci:slow test fix

* [run_slow] vivit
This commit is contained in:
Prakarsh Kaushik
2024-10-02 00:44:35 +05:30
committed by GitHub
parent 8635802af9
commit 68a2b50069
2 changed files with 7 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tests/models/vivit/test_modeling_vivit.py
View File

@@ -359,12 +359,12 @@ class VivitModelIntegrationTest(unittest.TestCase):
# allowing to interpolate the pre-trained position embeddings in order to use
# the model on higher resolutions. The DINO model by Facebook AI leverages this
# to visualize self-attention on higher resolution images.
model = VivitModel.from_pretrained("google/vivit-b-16x2").to(torch_device)
model = VivitModel.from_pretrained("google/vivit-b-16x2-kinetics400").to(torch_device)
image_processor = VivitImageProcessor.from_pretrained("google/vivit-b-16x2")
image_processor = VivitImageProcessor.from_pretrained("google/vivit-b-16x2-kinetics400")
video = prepare_video()
inputs = image_processor(
video, size={"shortest_edge": 480}, crop_size={"height": 480, "width": 480}, return_tensors="pt"
video, size={"shortest_edge": 480}, crop_size={"height": 232, "width": 232}, return_tensors="pt"
)
pixel_values = inputs.pixel_values.to(torch_device)