So I have a code that is much complex. Given that, I tried to recreate the issue using the official example of pytorch on nn.DataParallel. I am trying to have a forward function that goes two routes depending on the value of route variable. if route is True it does normal forwarding using a Linear function. the output of function is given to out which is an object of the Class. If route is False then I just ask it to print the shape of out. Surprisingly, the model thinks the out is still None, which is weird. So either I do not understand how DataParallel works or this is a serious issue. I appreciate any insight on this.
import torch
import torch.nn as nn
from torch.utils.data import Dataset, DataLoader
import os
os.environ["CUDA_VISIBLE_DEVICES"]="0, 1"
input_size = 5
output_size = 2
batch_size = 30
data_size = 100
class RandomDataset(Dataset):
def __init__(self, size, length):
self.len = length
self.data = torch.randn(length, size)
def __getitem__(self, index):
return self.data[index]
def __len__(self):
return self.len
rand_loader = DataLoader(dataset=RandomDataset(input_size, data_size),
batch_size=batch_size, shuffle=True)
class Model(nn.Module):
# Our model
def __init__(self, input_size, output_size):
super(Model, self).__init__()
self.fc = nn.Linear(input_size, output_size)
self.out = None
def forward(self, x):
input = x[0]
route = x[1]
if route:
output = self.fc(input)
if self.out is None:
self.out = output
else:
torch.cat((self.out, output), 1)
return output
else:
print(self.out.shape)
return self.out.shape
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model = Model(input_size, output_size)
if torch.cuda.device_count() > 1:
print("Let's use", torch.cuda.device_count(), "GPUs!")
# dim = 0 [30, xxx] -> [10, ...], [10, ...], [10, ...] on 3 GPUs
model = nn.DataParallel(model)
model.to(device)
for data in rand_loader:
input = data.to(device)
output = model((input, True))
print("Outside: input size", input.size(),
"output_size", output.size())
model((input, False))
Here is the error:
AttributeError: Caught AttributeError in replica 0 on device 0.
Original Traceback (most recent call last):
File "/usr/local/lib/python3.8/dist-packages/torch/nn/parallel/parallel_apply.py", line 61, in _worker
output = module(*input, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(*input, **kwargs)
File "/tmp/ipykernel_3007/2423608035.py", line 49, in forward
print(self.out.shape)
AttributeError: 'NoneType' object has no attribute 'shape'