TypeError: 'module' object is not callable

[songqi-github]

hi, when I was running this code:

import torch
import os
from torch2trt import torch2trt
from torchvision.models.resnet import resnet18

os.environ["CUDA_VISIBLE_DEVICES"]="2"
model = resnet18(pretrained=True).eval().cuda()
x = torch.ones((1, 3, 224, 224)).cuda()
model_trt = torch2trt(model, [x])
torch.save(model_trt.state_dict(), 'resnet18_trt.pth')

then, there comes out "TypeError: 'module' object is not callable"
how can I fix that?

[songqi-github]

waiting for your answer! THX!

[jaybdub]

Hi songqi-github,

Thanks for reaching out!

Do you witness this behavior without os.environ["CUDA_VISIBLE_DEVICES"]="2"?

Also, do you mind sharing

• PyTorch Version
• TensorRT version
• Torchvision Version
• Platform (ie: Jetson Xavier, etc.)

Best,
John

[yashraj02]

Hi even I am getting the same error any help highly appreciated!!

load net

cfg = widerface_640
num_classes = len(WIDERFace_CLASSES) + 1 # +1 background
net = build_ssd('test', cfg['min_dim'], num_classes) # initialize SSD
net.load_state_dict(torch.load(args.trained_model))

net.cuda()
net.eval()

import tensorrt as trt
data = torch.randn((1, 3, 640, 640))
data = torch.cuda.FloatTensor(data)
temp = torch2trt(net, [data])

Traceback (most recent call last):
File "widerface_val.py", line 249, in
temp = torch2trt(net, [data])
TypeError: 'module' object is not callable

[songqi-github]

Hi, I got the same error without os.environ["CUDA_VISIBLE_DEVICES"]="2".

PyTorch Version==1.3.0
TensorRT version==6.0.1.5
Torchvision Version==0.4.1

[songqi-github]

hi @yashraj02 Have you solved this problem?

[swpucl]

@songqi-github I get the same error.Do you solve it?

[sjtuherisson]

anyone use 'import torch2trt' replace 'from torch2trt import torch2trt' ?

[WeibinDai]

Hi, all. I meet the same problem, have anyone alreadly solved this issue?

[Tony-Yan2018]

Hi guys, I think what you should write is model_trt = torch2trt.torch2trt(model, [x]) instead of model_trt = torch2trt(model, [x]).
Hope this helps!

[OronG13]

@Tony-Yan2018
Yes, your answer help me and now it works well!
Thanks

[niranjanravilevelup]

Hi @jaybdub ,
I am facing the same error. I tried the above suggestion. It dint work. Any other alternatives?

[chainfitness]

@Tony-Yan2018 appreciate your advice, I tried it but the new error torch2trt has no attribute torch2trt was acquired. could you help me? thanks in advance

Hi guys, I think what you should write is model_trt = torch2trt.torch2trt(model, [x]) instead of model_trt = torch2trt(model, [x]). Hope this helps!

[chainfitness]

@sjtuherisson I used it replaced. but the 'module' object is not callable was acquired.

anyone use 'import torch2trt' replace 'from torch2trt import torch2trt' ?