att.config¶

Reproducibility infrastructure — seed management and YAML experiment configs.

att.config.set_seed(seed)[source]¶

Set global random seed for all stochastic operations.

Seeds NumPy, SciPy (via NumPy), and Python’s random module. Call once at the start of any experiment or notebook.

att.config.get_rng(seed=None)[source]¶

Get a numpy Generator, using provided seed or global state.

If seed is given, creates a new Generator from that seed. If seed is None, returns the global Generator (or creates one from seed=0).

att.config.load_config(path)[source]¶

Load experiment configuration from a YAML file.

Supported keys: seed, embedding, topology, binding, benchmarks, transitions, surrogates, system, dataset, preprocessing.

att.config.save_config(config, path)[source]¶

Save experiment parameters for reproducibility.

Parameters:

Return type:

None

Global seed management for deterministic reproducibility.

att.config.seed.set_seed(seed)[source]¶

Set global random seed for all stochastic operations.

Seeds NumPy, SciPy (via NumPy), and Python’s random module. Call once at the start of any experiment or notebook.

att.config.seed.get_rng(seed=None)[source]¶

Get a numpy Generator, using provided seed or global state.

If seed is given, creates a new Generator from that seed. If seed is None, returns the global Generator (or creates one from seed=0).

att.config.seed.get_seed()[source]¶

Return the current global seed, or None if not set.

YAML experiment configuration for reproducibility.