import torch
from typeguard import check_argument_types
[docs]class ForwardAdaptor(torch.nn.Module):
"""Wrapped module to parallelize specified method
torch.nn.DataParallel parallelizes only "forward()"
and, maybe, the method having the other name can't be applied
except for wrapping the module just like this class.
Examples:
>>> class A(torch.nn.Module):
... def foo(self, x):
... ...
>>> model = A()
>>> model = ForwardAdaptor(model, "foo")
>>> model = torch.nn.DataParallel(model, device_ids=[0, 1])
>>> x = torch.randn(2, 10)
>>> model(x)
"""
def __init__(self, module: torch.nn.Module, name: str):
assert check_argument_types()
super().__init__()
self.module = module
self.name = name
if not hasattr(module, name):
raise ValueError(f"{module} doesn't have {name}")
[docs] def forward(self, *args, **kwargs):
func = getattr(self.module, self.name)
return func(*args, **kwargs)