HTR Model Palatinus graecus 23 (Meleagre-NFD-finetuned)

HTR model for the Codex palatinus graecus 23 (Palatine Anthology), Byzantine Greek writing from the X^th century.

The model was trained from the ground truth produced by the Canada Research Chair on Digital Textualities, as part of the Anthologia graeca project. We focused our ground truth on 50 pages (143-195) and did finetuning on 20 extra pages (196-215).

This model was trained from the CREMMA Medieval model using NFD.

A segmentation model (meleagre-segment) was also trained with the 70 pages transcribed.