@@ -33,7 +33,9 @@ Sigma is always transcribed "σ" and never "ς". This decision reflects the scri
### Symbols
Every significant symbol was assigned a Unicode symbol using a correspondence table (see [table_CPgr23.csv](./data/CPgr23/table_CPgr23.csv) for every character used in our transcription).
Every significant symbol was assigned a Unicode symbol using a correspondence table (see [table_CPgr23.csv](./data/table_CPgr23.csv) for every character used in our transcription).
A [virtual keyboard](./data/AP.json) have been made using eScriptorium platform to help transcribing symbols.
These symbols have been chosen to be as similar as possible to the symbols used in the manuscript.
...
...
@@ -55,8 +57,12 @@ All abbreviations have been transcribed in expanded form: for example, "ȣ" is t
## Model description
<!-- to be completed -->
A transcription model for Greek manuscripts was trained using this dataset. It can be found here: {placeholder}.
3 transcription models for Greek manuscripts were trained using this dataset. They can be found here:
[HTR Model Palatinus graecus 23 (Meleagre-NFD-finetuned)](https://doi.org/10.5281/zenodo.10912725)
[HTR Model Palatinus graecus 23 (Meleagre-NFD)](https://doi.org/10.5281/zenodo.10912647)
[HTR Model Palatinus graecus 23 (Meleagre-NFC)](https://doi.org/10.5281/zenodo.10912498)
