aitune.torch.tune_strategy.one_backend_strategy

Simple tune strategy.

Module Contents

Classes

Name	Description
`OneBackendStrategy`	Strategy which uses just one provided backend.

API

class aitune.torch.tune_strategy.one_backend_strategy.OneBackendStrategy(
    backend: aitune.torch.backend.backend.Backend,
    kwargs = {}
)

Bases: PerformanceValidationMixin

Strategy which uses just one provided backend.

aitune.torch.tune_strategy.one_backend_strategy.OneBackendStrategy._describe_parts()

Describes the tuning.

aitune.torch.tune_strategy.one_backend_strategy.OneBackendStrategy._tune(
    module: torch.nn.Module,
    name: str,
    graph_spec: aitune.torch.module.graph_spec.GraphSpec,
    data: list[aitune.torch.module.recording_module.Sample],
    device: torch.device,
    cache_dir: pathlib.Path
) -> aitune.torch.backend.backend.Backend

Tunes given torch module with provided graph_spec and data.

aitune.torch.tune_strategy.one_backend_strategy.OneBackendStrategy.to_json_dict() -> dict[str, typing.Any]

Returns config dict for one backend strategy.