Here we tried three differenct methods for predicting price based on the title and description of the merchandise
TFIDF_model: MLP model based on the TFIDF vector representation of the items
TFIDF_model_with_interact: MLP model based on the TFIDF vector representation of the items + interaction between title and category of the item
GRU_model: use GRU to convert the item title and descriptions to vectors and build regression neural network on top of it.