Dp 627 en
Skočit na navigaci
Skočit na vyhledávání
Unstructured text comprehension and question answering
Author: Konrád Jakub
This thesis deals with comprehension of unstructured text and question answering. Specifically we focus the benefits of transfer learning. We show several experiments where we employ transfer learning to adjust model created on general large QA dataset (SQuAD) with data from smaller topic specific datasets. We explore influence of transfer learning on datasets focusing on specific topic not included in the main domain.
- Jakub Konrád, mailto:konrajak@fel.cvut.cz