Makes sense to quickly kick off with a simpler faster model. To so we need to - [ ] Improve the tokenizer - [ ] Store different vector types in the table (which entails different model types and a more relational way to think of predictions?) - [ ] Calcualte stop words and store in the db (make sure they update when a user adds data) - [ ] Account for vocab / vector size
Makes sense to quickly kick off with a simpler faster model.
To so we need to