Labeling data for handwritten text recognition (IAM/RIMES)

  huggingface-transformers, python-3.x, pytorch

I have a pre-trained model for recognition of handwritten text and I want to fine-tune it for my own data.
The input of the model is as prepared from an xml file as follow:

<?xml version='1.0' encoding='UTF-8'?> <DocumentList><SinglePage FileName="IAM/2010/0202/formsA-D/a01-000u.tif">
    <WriterID Name="000" />
    <PageClass Name="Press_reportage/political" />
    <Paragraph Bottom="1912" Left="363" Right="2414" Top="746" Value="A MOVE to stop Mr. Gaitskell from n nominating any more Labour life Peers n is to be made at a meeting of Labour n M Ps tomorrow . Mr. Michael Foot has n put down a resolution on the subject n and he is to be backed by Mr. Will n Griffiths , M P for Manchester Exchange . n ">
        <Line Bottom="835" Left="408" Right="2069" Top="746" Value="A MOVE to stop Mr. Gaitskell from">
            <Word Bottom="819" Left="408" Right="435" Top="768" Value="A" /><Word Bottom="814" Left="507" Right="720" Top="766" Value="MOVE" />
            <Word Bottom="814" Left="796" Right="866" Top="764" Value="to" />
            <Word Bottom="835" Left="919" Right="1085" Top="757" Value="stop" />
            <Word Bottom="815" Left="1185" Right="1311" Top="754" Value="Mr." />
            <Word Bottom="819" Left="1438" Right="1820" Top="746" Value="Gaitskell" />
            <Word Bottom="829" Left="1896" Right="2069" Top="757" Value="from" />

I want to know if anyone can help me with references of how to prepare my own data for this model.

I would really appreciate it!

Source: Python-3x Questions