Comment by rdedev

Ive been looking at a way to use transformer based models on tabular data. The hope is that these models have a much better contextual understanding of words. So embeddings from these models should be of better quality than just word2vec ones