Team Oasis colab notebooks and stuff for National Tech Olympycs - 2021

Typical text classification

• models used:

• https://huggingface.co/DeepPavlov/rubert-base-cased

• https://huggingface.co/xlm-roberta-base

• Dataset

  • https://russiansuperglue.com/tasks/download/MuSeRC

There are a lot of ways to handle this dataset. We tried (partially succesfull): QA, Classification, Span generation with and withous NER

• Dataset

  • https://russiansuperglue.com/tasks/download/RuCoS

• models (transforsmers only, obviously) used:

  • For text classification: https://huggingface.co/DeepPavlov/rubert-base-cased ,via slavic langs vocab provided by from DeepPavlov
  • For correct span generation: https://huggingface.co/google/mt5-large
  • For QA: https://huggingface.co/bert-base-multilingual-cased