wehub-resource-sync
44 lines
1.3 KiB
Python
44 lines
1.3 KiB
Python
import torch
|
|
|
|
|
|
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):
|
|
"""Initialize ForwardAdaptor.
|
|
|
|
Args:
|
|
module: TODO.
|
|
name: TODO.
|
|
"""
|
|
super().__init__()
|
|
self.module = module
|
|
self.name = name
|
|
if not hasattr(module, name):
|
|
raise ValueError(f"{module} doesn't have {name}")
|
|
|
|
def forward(self, *args, **kwargs):
|
|
"""Forward pass for training.
|
|
|
|
Args:
|
|
*args: Variable positional arguments.
|
|
**kwargs: Additional keyword arguments.
|
|
"""
|
|
func = getattr(self.module, self.name)
|
|
return func(*args, **kwargs)
|