Hi,
I am working on a project for table extraction to run on a firewalled laptop, hence am exploring downloading the pretrained model locally before transferring it over to run. However, I faced two main issues, specifically speaking with the table-transformers-detection model.
-
When I load the model locally (instead of accessing the hub for the model), the output class seems to be different and I keep encountering this error:
AttributeError: ‘TableTransformerModelOutput’ object has no attribute ‘logits’ -
Using the FeatureExtractor from the local loaded model seems to also provide different outputs as expected (I don’t face any errors, but loading locally instead of online) leads to different undesired results.
I am not extremely adept in this field and hence am unsure of whether it is due to some poor adjustment on my part, and would greatly appreciate any advice and suggestions!