It reduces the dimensionality of the input data

It allows the model to understand the semantic meaning of words

It enables the model to handle only numerical inputs

It speeds up the training process by skipping the embedding step