Darshan Gangurde

range of loss while training

3

what is overall reduction in loss for a good model? I am getting at best loss reduced to 3.7 starting from 4.7 at initial steps.. @naiveHobo

does it perform well on Name extraction?

1

Has anyone trained this model for extracting person name from pdfs? how was the result and how much data is sufficient for better performance?

Saving candidates ( fix #26 ) Please review @CS-savvy

Create candidates_to_json.py

1

Please review for relevance

Candidates directory

5

why files not found in candidates directory when training is started?

get_tesseract_results needs a path