Skip to content

CLIP : Enable CLIP.from_preset() Kaggle tests #2357

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.

Sign up for GitHub

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

Merged
merged 2 commits into from
Feb 23, 2024
Merged

CLIP : Enable CLIP.from_preset() Kaggle tests #2357

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e8000e8
update kaggle handle and weights conversion file
Feb 22, 2024
4093c8c
code reformat
Feb 22, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion keras_cv/models/feature_extractor/clip/clip_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,7 +72,7 @@ def __init__(
vision_patch_size=32,
context_length=77,
vocab_size=49408,
transformer_width=768,
transformer_width=512,
transformer_heads=8,
transformer_layers=12,
**kwargs,
Expand Down
10 changes: 5 additions & 5 deletions keras_cv/models/feature_extractor/clip/clip_model_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,9 +52,9 @@ def test_clip_model_golden_values(self):
processed_image, processed_text, attention_mask
)
print(image_logits)
self.assertAllClose(image_logits, [[2.932678, 2.932678, 2.932675]])
self.assertAllClose(image_logits, [[1.896713, 1.896713, 1.896713]])
self.assertAllClose(
text_logits, ops.transpose([[2.932678, 2.932678, 2.932675]])
text_logits, ops.transpose([[1.896713, 1.896713, 1.896713]])
)

def test_clip_preprocessor(self):
Expand All @@ -77,8 +77,8 @@ def test_clip_preprocessor_tf_data(self):

@pytest.mark.large
def test_presets(self):
self.skipTest("TODO: Enable after Kaggle model is public")
model = CLIP.from_preset("clip-vit-base-patch32")
# self.skipTest("TODO: Enable after Kaggle model is public")
model = CLIP.from_preset("clip-vit-base-patch16")
processed_image = np.ones(shape=[1, 224, 224, 3])
processed_text = np.ones(shape=[3, 77])
attention_mask = np.ones(shape=[3, 77])
Expand Down Expand Up @@ -109,7 +109,7 @@ def test_text_encoder_golden_values(self):
print(model.text_embeddings)
self.assertAllClose(
model.text_embeddings[0, :3],
[-0.018502, 0.000906, 0.020372],
[0.007531, -0.038361, -0.035686],
)

@pytest.mark.large # Saving is slow, so mark these large.
Expand Down
8 changes: 4 additions & 4 deletions keras_cv/models/feature_extractor/clip/clip_presets.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,7 +28,7 @@
"official_name": "CLIP",
"path": "clip",
},
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-base-patch16/2",
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-base-patch16/4",
},
"clip-vit-base-patch32": {
"metadata": {
Expand All @@ -44,7 +44,7 @@
"official_name": "CLIP",
"path": "clip",
},
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-base-patch32/2",
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-base-patch32/4",
},
"clip-vit-large-patch14": {
"metadata": {
Expand All @@ -60,7 +60,7 @@
"official_name": "CLIP",
"path": "clip",
},
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-large-patch14/2",
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-large-patch14/4",
},
"clip-vit-large-patch14-336": {
"metadata": {
Expand All @@ -76,6 +76,6 @@
"official_name": "CLIP",
"path": "clip",
},
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-large-patch14-336/2", # noqa: E501
"kaggle_handle": "kaggle://keras/clip/keras/clip-vit-large-patch14-336/4", # noqa: E501
},
}
Loading