This project explores playlist continuation using a fine-tuned language model trained on the Million Playlist Dataset. It combines semantic embeddings and similarity search to recommend tracks that match a given playlist theme.