Find here:

• dataset (incl. predictions of our compared models)
• training scripts and configs
• inference scripts and configs
• evaluation notebooks

• To use gpu support
  • please install the right torch version manually https://pytorch.org/get-started/locally/
  • Or adjust setup.py with the machtching torch version
  • Be sure that you have a c compiler installed

• pip install . without any torch backend. Manual installation of pytorch is needed
• pip install .[TORCH] with the cuda version defined in

• Start training a model using the gsap-ner-train script and a config file.
• With an optional second parameter you can set the gpu cores to use (e.g. "0,2")
• Example: gsap-ner-train train/configs/gsap-scideberta.yaml
• Feel free to write your own config files