Sentence Transformers

Sat, 18 Jul 2026 10:15:05 +0000

Definition

Sentence Transformers are extensions of traditional Transformer models (like BERT) fine-tuned to produce meaningful dense vector representations for entire sentences. Unlike standard token-level models, these architectures pool token embeddings to create a single sentence embedding that captures holistic semantic meaning. They are optimized using contrastive learning objectives to ensure that semantically similar sentences have vectors that are close together in the embedding space. This makes them highly effective for downstream tasks requiring semantic comparison.

Architectures on English AI Terms Dictionary

Sentence Transformers

Definition