I have a table consisting of around 50 key-value pairs in the real estate context. For instance "Year built" and "1994" or "Lot Size" and "4 acres". Based on this table, which is generated synthetically, I need to formulate a continuous text. GAN BERT will probably be part of that model. How many real data examples do I need? Are 200 sufficient? Overall, what are the steps to solve this problem?
Thanks! NLP Choppa