s444415/donut
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

test

Browse Source
This commit is contained in:
Michał Kozłowski 2022-12-17 10:44:55 +01:00
parent 296647793e
commit cc2e6541cf
1 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
donut-eval.py
View File

@ -15,15 +15,15 @@ from sconf import Config
def main(config):
image_size = [1920, 2560]
config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
config_vision.encoder.image_size = image_size # (height, width)
config_vision.decoder.max_length = 768
# image_size = [1920, 2560]
# config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
# config_vision.encoder.image_size = image_size # (height, width)
# config_vision.decoder.max_length = 768
processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config_vision)
model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path)
processor.image_processor.size = image_size[::-1] # should be (width, height)
# processor.image_processor.size = image_size[::-1] # should be (width, height)
processor.image_processor.do_align_long_axis = False
dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)