lilac.concepts: Concepts

Concepts are customizable signals that help enrich documents.

class lilac.concepts.ConceptUpdate

An update to a concept.

param insert: list[ExampleIn] | None = []

param remove: list[str] | None = []

param update: list[Example] | None = []

model_computed_fields: ClassVar[dict[str, ComputedFieldInfo]] = {}

A dictionary of computed field names and their corresponding ComputedFieldInfo objects.

class lilac.concepts.DiskConceptDB(project_dir: str | Path | None = None)

A concept database.

concept_acls(namespace: str, name: str, user: UserInfo | None = None) → ConceptACL

Return the ACL for a concept.

create(namespace: str, name: str, type: ConceptType | str = ConceptType.TEXT, metadata: ConceptMetadata | None = None, user: UserInfo | None = None) → Concept

Create a concept.

edit(namespace: str, name: str, change: ConceptUpdate, user: UserInfo | None = None) → Concept

Edit a concept. If the concept doesn’t exist, throw an error.

get(namespace: str, name: str, user: UserInfo | None = None) → Concept | None

Return a concept or None if there isn’t one.

list(user: UserInfo | None = None) → list[ConceptInfo]

List all the concepts.

merge_draft(namespace: str, name: str, draft: str, user: UserInfo | None = None) → Concept

Merge a draft concept.

namespace_acls(namespace: str, user: UserInfo | None = None) → ConceptNamespaceACL

Return the ACL for a namespace.

remove(namespace: str, name: str, user: UserInfo | None = None) → None

Remove a concept.

update_metadata(namespace: str, name: str, metadata: ConceptMetadata, user: UserInfo | None = None) → None

Update the metadata of a concept.

Parameters:

• namespace – The namespace of the concept.

• name – The name of the concept.

• metadata – The metadata to update.

• user – The user updating the metadata, if authentication is enabled.

class lilac.concepts.DiskConceptModelDB(concept_db: ConceptDB, project_dir: str | Path | None = None)

Interface for the concept model database.

create(namespace: str, concept_name: str, embedding_name: str, user: UserInfo | None = None) → ConceptModel

Create the concept model.

get(namespace: str, concept_name: str, embedding_name: str, user: UserInfo | None = None) → ConceptModel | None

Get the model associated with the provided concept the embedding.

Returns None if the model does not exist.

get_models(namespace: str, concept_name: str, user: UserInfo | None = None) → list[ConceptModel]

List all the models associated with a concept.

remove(namespace: str, concept_name: str, embedding_name: str, user: UserInfo | None = None) → None

Remove the model of a concept.

class lilac.concepts.Example

A single example in a concept used for training a concept model.

param id: str [Required]

model_computed_fields: ClassVar[dict[str, ComputedFieldInfo]] = {}

A dictionary of computed field names and their corresponding ComputedFieldInfo objects.

class lilac.concepts.ExampleIn

An example in a concept without the id (used for adding new examples).

param draft: str | None = 'main'

param img: bytes | None = None

param label: bool [Required]

param origin: ExampleOrigin | None = None

param text: str | None = None

model_computed_fields: ClassVar[dict[str, ComputedFieldInfo]] = {}

A dictionary of computed field names and their corresponding ComputedFieldInfo objects.