Size mismatch when loading CRNN weights in OCR model

[hajaraniwana]

Hi everyone,

I'm currently working on an OCR project using Doctr, and I encountered a size mismatch error when loading the weights for a CRNN model. Below is the relevant part of my code:

reco_model = crnn_vgg16_bn(pretrained=False, pretrained_backbone=False) reco_params = torch.load('/path/to/crnn_vgg16_bn_20241023-163747.pt', map_location="cpu") reco_model.load_state_dict(reco_params)

When I run the code, I get the following error:
RuntimeError: Error(s) in loading state_dict for CRNN: size mismatch for linear.weight: copying a param with shape torch.Size([127, 256]) from checkpoint, the shape in current model is torch.Size([124, 256]). size mismatch for linear.bias: copying a param with shape torch.Size([127]) from checkpoint, the shape in current model is torch.Size([124]).
It seems like there’s a size mismatch in the linear.weight and linear.bias layers when loading the checkpoint. I’m not sure why this is happening, as I expected the checkpoint to be compatible with the current model.

Does anyone have suggestions on how to resolve this size mismatch or insights into why this might be happening?

[hajaraniwana]

I have found the answared because vocab when i training, if any body have some issue refer to this solution link: #1109