aitune.torch.task.profiling.config | NVIDIA AITune Documentation

Profiling configuration.

Module Contents

Classes

Name	Description
`ProfilingConfig`	Configuration for profiling.

API

class aitune.torch.task.profiling.config.ProfilingConfig(
    batch_sizes: list[int] | collections.abc.Generator[int, None, None],
    batching: bool = True,
    measuring_strategy: aitune.torch.task.profiling.measuring_strategy.MeasuringStrategy | None = None,
    measurement_stop_strategy: aitune.torch.task.profiling.measuring_stop_strategy.MeasuringStopStrategy | None = None,
    profiling_stop_strategy: aitune.torch.task.profiling.profiling_stop_strategy.ProfilingStopStrategy | None = None
)

Dataclass

Configuration for profiling.

Note: all profiling configuration strategies are deepcopied before use.

batch_sizes

list[int] | Generator[int, None, None]

batching

bool = True

measurement_stop_strategy

MeasuringStopStrategy | None = None

measuring_strategy

MeasuringStrategy | None = None

profiling_stop_strategy

ProfilingStopStrategy | None = None

aitune.torch.task.profiling.config.ProfilingConfig.__post_init__()

Post-init.