Convert text into embeddings

An embedding model converts text into a numeric vector. Texts with related meaning are usually closer together in the embedding space, which makes embeddings useful for semantic search, RAG, classification, and recommendations.

Create an embedding model

import akasha.helper as ah

embedding_model = ah.handle_embeddings(
    "openai:text-embedding-3-small"
)

The provider alias selects the embedding integration. You must configure the corresponding provider credentials before making a remote request.

Use an Ollama embedding model

Akasha can use an embedding model served by Ollama. First make sure Ollama is running and pull the model:

ollama pull nomic-embed-text

Create the embedding object with an ollama: alias, then use the same embed_query() and embed_documents() methods:

import akasha.helper as ah

embedding_model = ah.handle_embeddings("ollama:nomic-embed-text")

query_vector = embedding_model.embed_query("What is retrieval?")
document_vectors = embedding_model.embed_documents(
    ["The first document discusses retrieval."]
)

print("Query dimensions:", len(query_vector))
print("Document dimensions:", len(document_vectors[0]))

Akasha connects to http://localhost:11434 by default. For another Ollama server, set OLLAMA_API_BASE or include the endpoint in the alias:

OLLAMA_API_BASE=http://192.168.1.10:11434

embedding_model = ah.handle_embeddings(
    "ollama:http://192.168.1.10:11434@nomic-embed-text"
)

Embed one query

vector = embedding_model.embed_query(
    "This is text that will be converted into an embedding."
)

print(type(vector))
print("Dimensions:", len(vector))
print("First values:", vector[:5])

embed_query() returns a list of floating-point values. The vector dimension depends on the selected embedding model.

Embed multiple documents

texts = [
    "The first document discusses retrieval.",
    "The second document discusses agents.",
    "The third document discusses memory.",
]

vectors = embedding_model.embed_documents(texts)

print("Documents:", len(vectors))
print("Dimensions:", len(vectors[0]))

embed_documents() returns one vector for each input text, in the same order as the input list.

Embeddings in RAG

You usually do not need to call the embedding methods directly for RAG. Pass an embedding alias to akasha.RAG() and Akasha handles document and query embeddings:

import akasha

rag = akasha.RAG(
    model="gemini:gemini-2.5-flash",
    embeddings="ollama:nomic-embed-text",
)

answer = rag("./docs", "What is the main topic?")
print(answer)

Warning

Remote embedding calls may incur provider charges. Do not commit private text, API keys, or generated vectors to a public repository.