List project contents via API

Python package: neptune-fetcher

To retrieve the names of runs, experiments, or attributes present in a project, use the list_*() methods of the Fetcher API.

Listing experiments

To list experiments in a project, use list_experiments().

To list only experiments whose name match a certain regex pattern:

import neptune_fetcher.alpha as npt


npt.list_experiments(r"exp_.*")

Output

['exp_xjjrq', 'exp_ymgun', 'exp_nmjbt', 'exp_ixgwm', 'exp_rdmuw']

Listing runs

To list runs in a project, use list_runs():

from neptune_fetcher.alpha import runs


runs.list_runs(runs=r"kittiwake_02.*25$")

Output

['onerous-kittiwake_0287625', 'spurious-kittiwake_025c425']

Listing unique attributes

To list attributes present in a project, use list_attributes().

The following example looks for attributes matching the regular expression metrics.*|config/.*, among experiments whose name match exp.*:

npt.list_attributes(
    experiments=r"exp.*",
    attributes=r"metrics.*|config/.*",
)

Output

['config/batch_size',
 'config/epochs',
 'config/last_event_time',
 'config/learning_rate',
 'config/optimizer',
 'config/use_bias',
 'metrics/accuracy',
 'metrics/loss',
 'metrics/val_accuracy',
 'metrics/val_loss']