Training and evaluation scripts used in my work "COMPRESSING BERT 25 TIMES BY RNN IN NAMED ENTITY RECOGNITION TASK" for student session of Dialog 2022 conference. Uses https://github.com/iluvvatar/research-ml-pipeline
Initially I didn't plan to publish this scripts, because they being changed while experiments and require refactoring. So after that work base library https://github.com/iluvvatar/research-ml-pipeline has been changed. I tried to adapt this scripts for last version on this moment but don't guarantee that all scripts work as expected.
To use:
pip install -r requirements.txt