FAQ

Frequently asked questions

Please add training related questions and answers here until they are added to the documentation.

I get "Failed to load any lstm-specific dictionaries for lang test!!" with my newly trained model

Newly trained models don't include a dictionary unless you add one. So add a dictionary to silence the warning or just ignore it.

Which tools can be used to generate ground truth for training?

These tools support transcriptions and produce results which can be used for training with tesstrain.

Small dedicated web applications

• archiscribe
• nashi
• ocr-gt-tools

More powerful tools

• Aletheia
• eScriptorium
• Transkribus