list_attributes()

Python package: neptune-query

Lists unique attributes in the runs of a Neptune project.

To limit the results, define filters for runs to search or attributes to include.

Related

• List attributes from experiments

Parameters

project

str

optional

default: None

Path of the Neptune project, as WorkspaceName/ProjectName.

If not provided, the NEPTUNE_PROJECT environment variable is used.

runs

str | list of str | Filter

optional

default: None

Filter specifying which runs to search.

• If a string is provided, it's treated as a regex pattern that the run IDs must match.
• If a list of strings is provided, it's treated as exact run IDs to match.
• To provide a more complex condition on an arbitrary attribute value, pass a Filter object.

attributes

str | list of str | AttributeFilter

optional

default: None

Filter specifying which attributes to include.

• If a string is provided, it's treated as a regex pattern that the attribute names must match.
• If a list of strings is provided, it's treated as exact attribute names to match.
• To provide a more complex condition, pass an AttributeFilter object.

Returns

list[str] – A list of attribute names in the runs of the Neptune project.

Example

List all attributes of a specific run:

import neptune_query.runs as nq_runs


nq_runs.list_attributes(
    runs=["prompt-wolf-20250605132116671-2g2r1"],  # run ID
)

List all attributes that begin with "metrics":

import neptune_query.runs as nq_runs


nq_runs.list_attributes(attributes=r"^metrics")

Search a specific project for runs with a learning rate less than 0.01 and return the logged attributes:

import neptune_query.runs as nq_runs
from neptune_query.filters import Filter


nq_runs.list_attributes(
    project="team-alpha/sandbox",
    runs=Filter.lt("config/lr", 0.01),
)