mnemosine.embed.embed

Kind: function

Signature

embed(storage, node, text=None)

Documentation

Compute and store an embedding for node.

Uses storage.embed_fn to turn text into a vector, validates that the result is a non-empty sequence of numbers, and stores it via mnemosine.vec.VectorStore.set_vector. When text is omitted the node's content is embedded.

Args:

• storage (Storage): The storage whose embed_fn is used.
• node (Node): The node to embed.
• text (str | None): Explicit text to embed. Defaults to
• node.content.

Returns:

• list of float: The computed embedding vector.

Raises:

• EmbeddingRequired: If storage.embed_fn is not set, or there is no
• text to embed.
• VectorError: If embed_fn returns something other than a
• non-empty sequence of numbers.

Example:

>>> db.embed_fn = lambda text: [1.0 if "storage" in text else 0.0, 0.0]
>>> from mnemosine.embed import embed
>>> embed(db, doc)
[1.0, 0.0]