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

Add support for fast image processing in image-pretraining example (#37021)

Browse Source 
* Add support for fast image processing in image-pretraining example

Fix typo: correct tuple formatting in IMAGE_PROCESSOR_MAPPING_NAMES

Signed-off-by: jafraustro <jaime.fraustro.valdez@intel.com>

* Use fast image processor by default

Co-authored-by: Pavel Iakubovskii <qubvel@gmail.com>
Signed-off-by: jafraustro <jaime.fraustro.valdez@intel.com>

---------

Signed-off-by: jafraustro <jaime.fraustro.valdez@intel.com>
Co-authored-by: Pavel Iakubovskii <qubvel@gmail.com>
This commit is contained in:
Jaime Fraustro
2025-04-03 06:26:46 -06:00
committed by GitHub
parent 34ccfebf32
commit afafb84b59
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/pytorch/image-pretraining/run_mim.py
View File

@@ -359,7 +359,7 @@ def main():
IMAGE_PROCESSOR_TYPES = { IMAGE_PROCESSOR_TYPES = {
conf.model_type: image_processor_class for conf, image_processor_class in IMAGE_PROCESSOR_MAPPING.items() conf.model_type: image_processor_class for conf, image_processor_class in IMAGE_PROCESSOR_MAPPING.items()
} }
image_processor = IMAGE_PROCESSOR_TYPES[model_args.model_type]() image_processor = IMAGE_PROCESSOR_TYPES[model_args.model_type][-1]()
# create model # create model
if model_args.model_name_or_path: if model_args.model_name_or_path:

4
src/transformers/models/auto/image_processing_auto.py
View File

@@ -56,7 +56,7 @@ else:
IMAGE_PROCESSOR_MAPPING_NAMES = OrderedDict( IMAGE_PROCESSOR_MAPPING_NAMES = OrderedDict(
[ [
("align", ("EfficientNetImageProcessor",)), ("align", ("EfficientNetImageProcessor",)),
("aria", ("AriaImageProcessor")), ("aria", ("AriaImageProcessor",)),
("beit", ("BeitImageProcessor",)), ("beit", ("BeitImageProcessor",)),
("bit", ("BitImageProcessor",)), ("bit", ("BitImageProcessor",)),
("blip", ("BlipImageProcessor", "BlipImageProcessorFast")), ("blip", ("BlipImageProcessor", "BlipImageProcessorFast")),
@@ -142,7 +142,7 @@ else:
("shieldgemma2", ("Gemma3ImageProcessor", "Gemma3ImageProcessorFast")), ("shieldgemma2", ("Gemma3ImageProcessor", "Gemma3ImageProcessorFast")),
("siglip", ("SiglipImageProcessor", "SiglipImageProcessorFast")), ("siglip", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
("siglip2", ("Siglip2ImageProcessor", "Siglip2ImageProcessorFast")), ("siglip2", ("Siglip2ImageProcessor", "Siglip2ImageProcessorFast")),
("superglue", "SuperGlueImageProcessor"), ("superglue", ("SuperGlueImageProcessor",)),
("swiftformer", ("ViTImageProcessor", "ViTImageProcessorFast")), ("swiftformer", ("ViTImageProcessor", "ViTImageProcessorFast")),
("swin", ("ViTImageProcessor", "ViTImageProcessorFast")), ("swin", ("ViTImageProcessor", "ViTImageProcessorFast")),
("swin2sr", ("Swin2SRImageProcessor",)), ("swin2sr", ("Swin2SRImageProcessor",)),