The baseline model uses 1990 rows of data, embeds using Word2Vec and trains on LSTM. The other models change one factor.
The labels mean the following: