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

additional config

Browse Source
This commit is contained in:
Michał Kozłowski 2022-12-17 10:29:25 +01:00
parent 3cbed8144e
commit bdb7f5ef7e
1 changed files with 4 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
donut-eval.py
View File

@ -15,14 +15,12 @@ from sconf import Config
def main(config): def main(config):
# max_length = 768 config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
# image_size = [1920, 2560] config_vision.encoder.image_size = [1920, 2560] # (height, width)
# config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path) config_vision.decoder.max_length = 768
# config_vision.encoder.image_size = image_size # (height, width)
# config_vision.decoder.max_length = max_length
processor = DonutProcessor.from_pretrained(config.pretrained_processor_path) processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path) model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config_vision)
dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split) dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)
device = "cuda" if torch.cuda.is_available() else "cpu" device = "cuda" if torch.cuda.is_available() else "cpu"
model.eval() model.eval()