mipcandy.sanity_check

Contents

`mipcandy.sanity_check`#

Module Contents#

Classes#

SanityCheckResult

Functions#

`num_trainable_params`
`model_complexity_info`
`sanity_check`

API#

mipcandy.sanity_check.num_trainable_params(model: torch.nn.Module) → int[source]#

mipcandy.sanity_check.model_complexity_info(model: torch.nn.Module, example_shape: Sequence[int]) → tuple[float | None, float | None, str][source]#

class mipcandy.sanity_check.SanityCheckResult[source]#

Bases: object

num_macs: float = None#

num_params: float = None#

layer_stats: str = None#

output: torch.Tensor = None#

__str__() → str[source]#

mipcandy.sanity_check.sanity_check(model: torch.nn.Module, input_shape: Sequence[int], *, device: mipcandy.types.Device | None = None) → mipcandy.sanity_check.SanityCheckResult[source]#