AttributeFilter

Python package: neptune-query

Specifies criteria for attributes when using a fetching method.

Parameters

name

str | list[str]

optional

default: None

Criterion for attribute names.

If a string is provided, it's treated as a regex pattern that the attribute name must match. Supports Neptune's extended regex syntax.

If a list of strings is provided, it's treated as exact attribute names to match.

type

"bool" | "datetime" | "file" | "float" | "int" | "string" | "string_set" | "float_series" | "histogram_series" | "string_series" | "file_series"

optional

default: all available types

Allowed attribute types, as a literal string or a list of literal strings.

Examples

From a particular experiment, fetch values from all FloatSeries attributes with "loss" in the name:

import neptune_query as nq
from neptune_query.filters import AttributeFilter


losses = AttributeFilter(name=r"loss", type="float_series")
loss_values = nq.fetch_metrics(
    experiments=["training-week-34"],
    attributes=losses,
)