When using document understanding, there are multiple predefined models available. There seems to be no base model of DocPath when creating a document understanding project.
Are the predefined models based on DocPath?
What is the best method currently available to have better accuracy in extracting certain details from pdf/image document? Do I need to train on my data like invoices to get best result, or DocPath being an LLM has been trained on huge dataset already.
This link suggests almost all the models are based on DocPath, but does not explicitly lists them: