`AutoMAXConfigration`

class AutoMAXConfigration(**kwargs)

Configuration container for an AutoMAX hyperparameter-optimisation run.

All arguments are keyword-only and are consumed via **kwargs. Maps directly to the automax: section of the YAML config file.

Parameters:

deterministic (bool) – Enforce deterministic behaviour across trials.
n_trials (int) – Total number of HPO trials to run.
n_configs (int) – Number of configurations evaluated per trial.
SEED (int) – Random seed used by the SMAC3 sampler (default: 0).
name (str) – Human-readable identifier for the search run.
output_directory (str) – Directory where AutoMAX writes results (default: "./automax_output").
overwrite (bool) – Overwrite an existing output directory with the same name.

Example

from src.core import AutoMAXConfigration

automax_cfg = AutoMAXConfigration(
    deterministic=True,
    n_trials=50,
    n_configs=10,
    SEED=0,
    name="my_search",
    output_directory="./automax_output",
    overwrite=False,
)

YAML equivalent

automax:
  deterministic: true
  n_trials: 50
  n_configs: 10
  SEED: 0
  name: my_search
  output_directory: ./automax_output
  overwrite: false

Attributes

Attribute	Type	Description
`deterministic`	bool	Enforce deterministic training across all trials.
`n_trials`	int	Total number of HPO trials to run.
`n_configs`	int	Configurations evaluated per trial.
`SEED`	int	Random seed for the SMAC3 sampler (default: `0`).
`name`	str	Unique run identifier; also used as the SMAC3 scenario name.
`output_directory`	str	Root output directory (default: `"./automax_output"`).
`overwrite`	bool	Whether to overwrite an existing run with the same name.

AutoMAXConfigration

`AutoMAXConfigration`