search

clovaai / donut

Official Implementation of OCR-free Document Understanding Transformer (Donut) and Synthetic Document Generator (SynthDoG), ECCV 2022
https://arxiv.org/abs/2111.15664
MIT License
5.75k stars 466 forks source link

Error while training #172

Open rachiteagles opened 1 year ago

rachiteagles commented 1 year ago

I was training the model on a different dataset and image size i used was 250*250. I am getting error when i start training

RuntimeError: The size of tensor a (64) must match the size of tensor b (100) at non-singleton dimension 3

image

lauraminkova commented 1 year ago

Hi ! I am getting the same error, did you happen to figure out the problem ?

bikashg commented 9 months ago

Me as well, having the same issue when using image_size of 256 by 256. However, using a different image_size for the same dataset (specifying reconstruction on the feature extractor and encoder config), e.g. 480 by 360 works fine.