This adds an option add_i_node_text to the method convert_to_document() in dataset_builders/pie/dialam2024/dialam2024.py which allows us to combine the text of both L- and I-nodes.
It is set by default to False since the evaluation with the official script (see https://github.com/ArneBinder/dialam-2024-shared-task/pull/34) resulted in slightly worse scores compared to the original version that uses only L-node texts.
This adds an option
add_i_node_text
to the methodconvert_to_document()
indataset_builders/pie/dialam2024/dialam2024.py
which allows us to combine the text of both L- and I-nodes. It is set by default to False since the evaluation with the official script (see https://github.com/ArneBinder/dialam-2024-shared-task/pull/34) resulted in slightly worse scores compared to the original version that uses only L-node texts.