← Back to all frameworks NLP

Sentence-Transformers

Sentence-level embeddings for semantic search

What it is

A wrapper around BERT-style models specialized for producing fixed-size sentence embeddings. The standard for building dense retrieval, clustering, deduplication and semantic similarity.

How Vaaani uses it

  • Vector indexes for RAG (paired with Pinecone, Qdrant or Weaviate)
  • De-duplicating support tickets and FAQs
  • Building 'similar items' / recommendation features
  • Multilingual semantic search across mixed-script corpora

Why it makes the cut

Cheaper than calling OpenAI embeddings on millions of docs, and you control the model. The default embedding layer for self-hosted Vaaani RAG.

Sample code

from sentence_transformers import SentenceTransformer

m = SentenceTransformer("all-mpnet-base-v2")
emb = m.encode(["Vaaani builds AI workers",
                "Custom chatbots and automation"])
# 768-dim vectors, ready for Pinecone

Related in the Vaaani stack

Have a project that needs Sentence-Transformers?

30-min discovery call. You describe the busywork; I map it to an AI worker and a budget.