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

allow to use SDPA in clip models #941

Merged
merged 1 commit into from
Oct 17, 2024
Merged

allow to use SDPA in clip models #941

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
49 changes: 49 additions & 0 deletions optimum/exporters/openvino/model_configs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,8 @@
from optimum.exporters.onnx.model_configs import (
CLIPOnnxConfig,
CLIPTextOnnxConfig,
CLIPTextWithProjectionOnnxConfig,
CLIPVisionModelOnnxConfig,
CodeGenOnnxConfig,
FalconOnnxConfig,
GemmaOnnxConfig,
Expand All @@ -35,6 +37,7 @@
PhiOnnxConfig,
VisionOnnxConfig,
)
from optimum.exporters.onnx.model_patcher import ModelPatcher
from optimum.exporters.tasks import TasksManager
from optimum.utils import DEFAULT_DUMMY_SHAPES
from optimum.utils.input_generators import (
Expand Down Expand Up @@ -1079,6 +1082,11 @@ def generate_dummy_inputs_for_validation(
reference_model_inputs["text"] = reference_model_inputs.pop("input_ids")
return super().generate_dummy_inputs_for_validation(reference_model_inputs)

def patch_model_for_export(
self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
) -> ModelPatcher:
return ModelPatcher(self, model, model_kwargs=model_kwargs)


@register_in_tasks_manager("clip-text-model", *["feature-extraction"], library_name="open_clip")
class OpenCLIPTextOpenVINOConfig(CLIPTextOnnxConfig):
Expand Down Expand Up @@ -1109,6 +1117,11 @@ def generate_dummy_inputs(self, framework: str = "pt", **kwargs):
dummy_inputs = super().generate_dummy_inputs(framework=framework, **kwargs)
return dummy_inputs

def patch_model_for_export(
self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
) -> ModelPatcher:
return ModelPatcher(self, model, model_kwargs=model_kwargs)


@register_in_tasks_manager("clip-vision-model", *["feature-extraction"], library_name="open_clip")
class OpenCLIPVisualOpenVINOConfig(VisionOnnxConfig):
Expand All @@ -1134,6 +1147,42 @@ def rename_ambiguous_inputs(self, inputs):
return model_inputs


@register_in_tasks_manager(
"clip", *["feature-extraction", "zero-shot-image-classification"], library_name="transformers"
)
class CLIPOpenVINOConfig(CLIPOnnxConfig):
def patch_model_for_export(
self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
) -> ModelPatcher:
return ModelPatcher(self, model, model_kwargs=model_kwargs)


@register_in_tasks_manager("clip-text-model", *["feature-extraction"], library_name="transformers")
@register_in_tasks_manager("clip-text-model", *["feature-extraction"], library_name="diffusers")
class CLIPTextOpenVINOConfig(CLIPTextOnnxConfig):
def patch_model_for_export(
self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
) -> ModelPatcher:
return ModelPatcher(self, model, model_kwargs=model_kwargs)


@register_in_tasks_manager("clip-text-with-projection", *["feature-extraction"], library_name="transformers")
@register_in_tasks_manager("clip-text-with-projection", *["feature-extraction"], library_name="diffusers")
class CLIPTextWithProjectionOpenVINOConfig(CLIPTextWithProjectionOnnxConfig):
def patch_model_for_export(
self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
) -> ModelPatcher:
return ModelPatcher(self, model, model_kwargs=model_kwargs)


@register_in_tasks_manager("clip-vision-model", *["feature-extraction"], library_name="transformers")
class CLIPVisionModelOpenVINOConfig(CLIPVisionModelOnnxConfig):
def patch_model_for_export(
self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
) -> ModelPatcher:
return ModelPatcher(self, model, model_kwargs=model_kwargs)


@register_in_tasks_manager(
"ibert",
*[
Expand Down
Loading