A comparative study of deep learning based language representation learning models

Deep learning (DL) approaches use various processing layers to learn hierarchical representations of data. Recently, many methods and designs natural language (NLP) models have shown significant development, especially in text mining analysis. For vector-space text, there are famous like Word2vec, GloVe, fastText. In fact, NLP took a big step forward when BERT recently GTP-3 came out. this paper, we highlight the most important representation provide an insight their evolution. We also summarize, compare contrast these different on sentiment analysis, thus discuss main strengths limitations. Our obtained results show that is best model.