More details in Tutorial

From Section 10.2,

• The corpus contains 471 scanned documents pages related to Casimir Marius PELLET, a French soldier during World War I.
• Explain the project description supports Markdown input
• explain why transcription in Europeana is not directly usable to train a model
• three eval_* folders : explain why testing on trainset
• check that results.tar.gz can be opened on windows or link to tools how to open it