4 ms·

Could you please provide pointers to relevant papers? I'm very interested.

• desku 6y ago
  Here's a paper on how BERT (a large Transformer model trained using self-supervised learning) implicitly learns the traditional NLP pipeline: https://arxiv.org/abs/1905.05950 https://arxiv.org/abs/1905.05950