Nomic Embed
Open embedding model with an auditable training set and long context.
Overview
Nomic Embed is an open embedding model released with an open training dataset and reproducible training run, as well as long-context support. It solves two problems at once: most open embedding models publish weights but not the data they were trained on, which makes their behaviour hard to reason about, and most cap inputs short of a document. It exists because an embedding model whose training data is inspectable can be evaluated for the biases that come from that data.
Technology Graph
How Nomic Embed connects to the rest of the stack.
Applied to
- RAG technique Techniques Nomic Embed used for RAG High ✓ source
Alternatives
-
BGE
model
An alternative to Nomic Embed · source
-
Cohere Embed
model
An alternative to Nomic Embed · source
-
E5
model
An alternative to Nomic Embed · source
-
GTE
model
An alternative to Nomic Embed · source
-
Instructor
model
An alternative to Nomic Embed · source
-
Jina Embeddings
model
An alternative to Nomic Embed · source
-
OpenAI Text Embedding 3
model
An alternative to Nomic Embed · source
Comparisons
Side-by-side breakdowns with the alternatives above.