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.

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

[CLIPSeg] Make interpolate_pos_encoding default to True #34419

Merged
merged 6 commits into from
Oct 31, 2024
Merged

[CLIPSeg] Make interpolate_pos_encoding default to True #34419

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
91816dd
Remove interpolate_pos_encoding
NielsRogge Oct 25, 2024
33936e7
Make fixup
NielsRogge Oct 25, 2024
df982a5
Make interpolate_pos_encoding default to True
NielsRogge Oct 27, 2024
8fc4a33
Reuse existing interpolation
NielsRogge Oct 29, 2024
c121f4b
Add integration test
NielsRogge Oct 31, 2024
45c0f97
Merge branch 'main' into feature/undo_32600
NielsRogge Oct 31, 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
22 changes: 9 additions & 13 deletions src/transformers/models/clipseg/modeling_clipseg.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -205,7 +205,7 @@ def interpolate_pos_encoding(self, embeddings: torch.Tensor, height: int, width:

return torch.cat((class_pos_embed, patch_pos_embed), dim=1)

def forward(self, pixel_values: torch.FloatTensor, interpolate_pos_encoding=False) -> torch.Tensor:
NielsRogge marked this conversation as resolved.
Show resolved Hide resolved
def forward(self, pixel_values: torch.FloatTensor, interpolate_pos_encoding=True) -> torch.Tensor:
batch_size, _, height, width = pixel_values.shape
if not interpolate_pos_encoding and (height != self.image_size or width != self.image_size):
raise ValueError(
Expand Down Expand Up @@ -535,7 +535,7 @@ def _init_weights(self, module):
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
interpolate_pos_encoding (`bool`, *optional*, defaults to `False`):
interpolate_pos_encoding (`bool`, *optional*, defaults to `True`):
Whether to interpolate the pre-trained position encodings.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
Expand Down Expand Up @@ -574,7 +574,7 @@ def _init_weights(self, module):
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
interpolate_pos_encoding (`bool`, *optional*, defaults to `False`):
interpolate_pos_encoding (`bool`, *optional*, defaults to `True`):
Whether to interpolate the pre-trained position encodings.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
Expand Down Expand Up @@ -845,14 +845,13 @@ def __init__(self, config: CLIPSegVisionConfig):

@add_start_docstrings_to_model_forward(CLIPSEG_VISION_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=BaseModelOutputWithPooling, config_class=CLIPSegVisionConfig)
# Copied from transformers.models.clip.modeling_clip.CLIPVisionTransformer.forward
def forward(
self,
pixel_values: Optional[torch.FloatTensor] = None,
pixel_values: Optional[torch.FloatTensor],
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
interpolate_pos_encoding: Optional[bool] = False,
interpolate_pos_encoding: Optional[bool] = True,
) -> Union[Tuple, BaseModelOutputWithPooling]:
r"""
Returns:
Expand All @@ -864,9 +863,6 @@ def forward(
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict

if pixel_values is None:
raise ValueError("You have to specify pixel_values")

hidden_states = self.embeddings(pixel_values, interpolate_pos_encoding=interpolate_pos_encoding)
hidden_states = self.pre_layrnorm(hidden_states)

Expand Down Expand Up @@ -912,7 +908,7 @@ def forward(
pixel_values: Optional[torch.FloatTensor] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
interpolate_pos_encoding: Optional[bool] = False,
interpolate_pos_encoding: Optional[bool] = True,
return_dict: Optional[bool] = None,
) -> Union[Tuple, BaseModelOutputWithPooling]:
r"""
Expand Down Expand Up @@ -1035,7 +1031,7 @@ def get_image_features(
pixel_values: Optional[torch.FloatTensor] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
interpolate_pos_encoding: bool = False,
interpolate_pos_encoding: bool = True,
return_dict: Optional[bool] = None,
) -> torch.FloatTensor:
r"""
Expand Down Expand Up @@ -1091,7 +1087,7 @@ def forward(
return_loss: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
interpolate_pos_encoding: bool = False,
interpolate_pos_encoding: bool = True,
return_dict: Optional[bool] = None,
) -> Union[Tuple, CLIPSegOutput]:
r"""
Expand Down Expand Up @@ -1397,7 +1393,7 @@ def forward(
labels: Optional[torch.LongTensor] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
interpolate_pos_encoding: bool = False,
interpolate_pos_encoding: bool = True,
return_dict: Optional[bool] = None,
) -> Union[Tuple, CLIPSegOutput]:
r"""
Expand Down
4 changes: 2 additions & 2 deletions tests/models/clipseg/test_modeling_clipseg.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -796,15 +796,15 @@ def test_inference_image_segmentation(self):

# forward pass
with torch.no_grad():
outputs = model(**inputs, interpolate_pos_encoding=True)
outputs = model(**inputs)

# verify the predicted masks
self.assertEqual(
outputs.logits.shape,
torch.Size((3, 352, 352)),
)
expected_masks_slice = torch.tensor(
[[-7.4613, -7.4785, -7.3627], [-7.3268, -7.0898, -7.1333], [-6.9838, -6.7900, -6.8913]]
[[-7.4613, -7.4785, -7.3628], [-7.3268, -7.0899, -7.1333], [-6.9838, -6.7900, -6.8913]]
).to(torch_device)

self.assertTrue(torch.allclose(outputs.logits[0, :3, :3], expected_masks_slice, atol=1e-3))
Expand Down
Loading