For a broader introduction to the field, these resources are also highly recommended:
: This survey focusing on practical use explores open-access tools and real-world implementations, specifically where text is the primary modality.
: A high-level overview detailing how transformers became the go-to architecture not just for NLP, but also for computer vision and audio processing.
An essential paper for anyone starting out is by Tong Xiao and Jingbo Zhu. It serves as a comprehensive 119-page guide that bridges the gap between basic concepts and recent advanced techniques.
: A systematic review from 2024 that highlights how these models solve various NLP problems across different languages and domains.
: A 2023 review that demystifies the architecture by breaking it down into its core components for beginners.