Reference

Bases: BaseModel

An LLM represents an OpenAI-compatible LLM endpoint.

No validation takes place when creating or updating LLM objects in the database. The first time the server attempts a connection is when a query is performed.

Attributes:

Name	Type	Description
name	str	The identifier for the LLM object.
endpoint	str	OpenAI-compatible endpoint.
api_key	str | None	Secret API Key to connect to the LLM endpoint.

__call__(**params)

Return a LangChain Chat instance.

openai_chat_model(**params)

Return a LangChain OpenAI chat model.

Bases: BaseModel

An Embedding Provider represents a model the server supports.

Bases: BaseModel

An Embedding Model is a particular instantiation of an Embedding Provider with specific parameters.

__call__()

Return the associated LangChain Embedding instance.

Bases: BaseModel

An Text Splitter represents a provider the server supports.

__call__()

Return the associated LangChain Embedding instance.

Bases: BaseModel

A Pipeline contains all the logic and configuration on how to process RAG requests.

embedding_model()

Associated Embedding Model.

Bases: BaseModel

Class representing a document to be processed by a Pipeline.

Bases: BaseModel

An Agent represents a set of instructions with an associated pipeline and LLM.

Bases: BaseStore, VectorStore

Foundation4ai Vector Store.

from_texts(texts, embedding, embedding_length, metadatas=None, *, db, table, distance_strategy=DistanceStrategy.COSINE, ids=None, **kwargs) classmethod

Return VectorStore initialized from texts and embeddings.

Parameters:

Name	Type	Description	Default
texts	list[str]	Texts to add to the vectorstore.	required
embedding	Embeddings	Embedding function to use.	required
embedding_length	int	Embedding length	required
metadatas	Optional[List[dict]]	Optional list of metadatas associated with the texts. Default is None.	None
ids	Optional[List[str]]	Optional list of IDs associated with the texts.	None
db	Connection	database connection,	required
table	str	table name to save the vectors,	required
distance_strategy	DistanceStrategy	distance strategy for retrievals,	COSINE
kwargs	Any	Additional keyword arguments.	{}

Returns:

Name	Type	Description
VectorStore	VST	VectorStore initialized from texts and embeddings.

Bases: BaseModel

Classification search parameters.

Bases: IntFlag

Type of classifications search.

EXACT = 1 class-attribute instance-attribute

Causes the search to fail if any classification passed in doesn't exist in the pipeline.

HIERARCHICAL = 2 class-attribute instance-attribute

Search is hierarchical

OpenAI Support

Chat model.