Getters

EncryptedIndex exposes index settings as properties, plus two training-status methods. Every getter except index_name calls the service’s describe or training-status endpoint. On HTTP errors, getters raise the same CyborgDBError subclasses as the other methods, for example NotFoundError if the index was deleted.

index_name

@property
index_name: str

The index’s name. Read locally, with no network call.

print(index.index_name)
# Output: my-vectors

dimension

@property
dimension: int

Vector dimensionality. Returns 0 if the index was created without a dimension and nothing has been upserted yet. Once non-zero, the value is cached and later reads don’t call the service.

print(index.dimension)
# Output: 384

metric

@property
metric: str

Distance metric: "euclidean", "squared_euclidean", or "cosine". Cached after the first read.

print(index.metric)
# Output: euclidean

n_lists

@property
n_lists: int

Number of IVF lists. 1 for an untrained index; the trained list count after train(). Fetched from the service on every read.

print(index.n_lists)
# Output: 1

metadata_schema

@property
metadata_schema: Dict[str, Dict[str, bool]]

The per-field indexing policy set with metadata_schema or text_fields in create_index(). Each entry has filterable, pattern, and full_text. {} when the index uses the default policy. Fetched on every read.

print(index.metadata_schema)
# Output: {'body': {'filterable': False, 'pattern': False, 'full_text': True}}

bm25

@property
bm25: Optional[Dict[str, Any]]

The index’s BM25 settings, or None when it has no full-text fields.

Key	Type	Description
`k1`	`float`	Term-frequency saturation.
`b`	`float`	Length normalization.
`analyzer_version`	`str` or `None`	Text analyzer the index was built with.

print(index.bm25)
# Output: {'k1': 1.2, 'b': 0.75, 'analyzer_version': 'en/1;sw=2de831ef7be676b6;snowball-english-utf8-3.0.1'}

is_trained

index.is_trained() -> bool

True if the index has been trained. Raises a CyborgDBError subclass if the describe request fails, for example NotFoundError if the index was deleted.

if not index.is_trained():
    index.train()

is_training

index.is_training() -> bool

True while the service is training this index or has it queued for training. Requires the root key: with a per-user key the service returns HTTP 403 and the SDK raises cyborgdb.AuthenticationError.

import time

while index.is_training():
    time.sleep(5)
print(index.is_trained())

Introduction

Client

Encrypted Index

Types & Helpers

index_name

dimension

metric

n_lists

metadata_schema

bm25

is_trained

is_training

​index_name

​dimension

​metric

​n_lists

​metadata_schema

​bm25

​is_trained

​is_training

index_name

dimension

metric

n_lists

metadata_schema

bm25

is_trained

is_training