Skip to content

Util

Utility functions for trial configuration key mapping and ID assignment.

get_config_key_to_id_mapping #

get_config_key_to_id_mapping(
    table: DataFrame, fid_name: str
) -> dict[tuple[tuple[str, Any], ...], int]

Map each unique trial configuration (excluding fidelity) to its highest config ID.

PARAMETER DESCRIPTION
table

DataFrame with trial configurations and IDs.

TYPE: DataFrame

fid_name

Name of the fidelity parameter to exclude.

TYPE: str

RETURNS DESCRIPTION
dict[tuple[tuple[str, Any], ...], int]

Dictionary mapping unique config keys to their highest config ID.
RAISES DESCRIPTION
TypeError

If table is not a pandas DataFrame.
Source code in neps\optimizers\utils\util.py 
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
def get_config_key_to_id_mapping(
    table: pd.DataFrame, fid_name: str
) -> dict[tuple[tuple[str, Any], ...], int]:
    """
    Map each unique trial configuration (excluding fidelity) to its highest config ID.

    Args:
        table: DataFrame with trial configurations and IDs.
        fid_name: Name of the fidelity parameter to exclude.

    Returns:
        Dictionary mapping unique config keys to their highest config ID.

    Raises:
        TypeError: If table is not a pandas DataFrame.
    """
    if not isinstance(table, pd.DataFrame):
        raise TypeError("table must be a pandas DataFrame.")
    config_to_id = {}
    for idx, row in table.iterrows():
        config_id = idx[0]
        config = row["config"]
        config_key = get_trial_config_unique_key(config, fid_name)
        # we always want to keep the highest config_id for a given config_key
        config_to_id[config_key] = config_id
    return config_to_id

get_trial_config_unique_key #

get_trial_config_unique_key(
    config: Mapping[str, Any], fid_name: str | None = None
) -> tuple[tuple[str, Any], ...]

Generate a unique key for a trial configuration, optionally excluding the fidelity parameter.

PARAMETER DESCRIPTION
config

Mapping of configuration parameters.

TYPE: Mapping[str, Any]

fid_name

Name of the fidelity parameter to exclude (if any).

TYPE: str | None DEFAULT: None

RETURNS DESCRIPTION
tuple[tuple[str, Any], ...]

A tuple of (key, value) pairs sorted by key, excluding the fidelity parameter if specified.
RAISES DESCRIPTION
TypeError

If config is not a mapping.
Source code in neps\optimizers\utils\util.py 
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
def get_trial_config_unique_key(
    config: Mapping[str, Any], fid_name: str | None = None
) -> tuple[tuple[str, Any], ...]:
    """
    Generate a unique key for a trial configuration,
    optionally excluding the fidelity parameter.

    Args:
        config: Mapping of configuration parameters.
        fid_name: Name of the fidelity parameter to exclude (if any).

    Returns:
        A tuple of (key, value) pairs sorted by key,
        excluding the fidelity parameter if specified.

    Raises:
        TypeError: If config is not a mapping.
    """
    if not isinstance(config, Mapping):
        raise TypeError("config must be a mapping (e.g., dict).")
    return tuple(
        sorted((k, v) for k, v in config.items() if fid_name is None or k != fid_name)
    )