7: Deep Learning for Natural Language – Transformers
MIT OpenCourseWare · 76:37
Transformers dominate modern ML because they produce same-length contextual representations that respect both surrounding words and word order; this lecture builds that architecture from first principles on ATIS fligh...