Learn&Implement: Additional datasets

With Jonas' return letter, we confirmed that we could use additional datasets to strengthen our model. In fact, the following two papers are recommended in the project description. I will try to implement them next.

SMART: Robust and Efficient Fine-Tuning for Pre-trained Natural Language Models through Principled Regularized Optimization. Jiang et al. [2020]

Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks. Reimers and Gurevych [2019]