🐱 Github: github.com/itsadeepizza/nlp

Language is probably one of the finest and most impressive inventions.

Not only does it serve as the primary means of conveying knowledge to others, but it also plays a crucial role in shaping our thoughts and understanding of the world around us.

The close relationship between language and thought is likely the reason why natural language processing (NLP) has been a driving force in the advancement of many techniques used in other areas of deep learning, such as CNNs, recurrent networks, and transfomers.

We worked on three different NLP projects:

  • The creation of a dataset sourced from the Italian version of Wikipedia, which was used to train a word2vec model for word embedding.

    🐱 Link github

  • The development of a custom transformer encoder from scratch, which was used to perform sentiment analysis on Italian tweets. Performance has been compared with a random forest classifier.

    🐱 Link github

  • The construction of a transformer decoder-only for text generation, which was trained using the Wikipedia dataset from our previous project.

    🐱 Link github