Splicing dependency maps

[pabloacera]

Hi,

Thank you for providing the code. I have a quick clarification regarding the generation of dependency maps using SpliceBERT, both in your paper and the accompanying Jupyter notebooks. Specifically, could you clarify:

Which SpliceBERT model was utilized to generate the dependency maps presented in the paper and notebooks?
Was the model fine-tuned specifically for generating dependency maps, or was it used directly as provided in the official SpliceBERT GitHub repository?

These points were not entirely clear in the preprint, so any additional details would be greatly appreciated.

Thanks again for your time!