Skip-gram

Algorithm that learns word embeddings by pushing each center word's vector toward the vectors of its neighbors in the corpus.

Companion explanation in Step by Token, chapter 3.