Tensorboardx: tensorboard does not show graph

[donttal]

Describe the bug
I use a simple example to add a linear model by add_graph(), generate "runs/Feb06_11-31-24_casperMac.localNet1/events.out.tfevents.1580959884.casperMac.local" file. but after I run the command "tensorboard --logdir=runs". Then I can not see the graph at http://casperMac.local:6006

Minimal runnable code to reproduce the behavior

# -*- coding: utf-8 -*-
import torch
import torch.nn as nn
import torch.nn.functional as F
from tensorboardX import SummaryWriter


class Net1(nn.Module):
    def __init__(self):
        super(Net1, self).__init__()
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
        self.conv2_drop = nn.Dropout2d()
        self.fc1 = nn.Linear(320, 50)
        self.fc2 = nn.Linear(50, 10)
        self.bn = nn.BatchNorm2d(20)

    def forward(self, x):
        x = F.max_pool2d(self.conv1(x), 2)
        x = F.relu(x) + F.relu(-x)
        x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))
        x = self.bn(x)
        x = x.view(-1, 320)
        x = F.relu(self.fc1(x))
        x = F.dropout(x, training=self.training)
        x = self.fc2(x)
        x = F.softmax(x, dim=1)
        return x


dummy_input = torch.rand(13, 1, 28, 28)

model = Net1()
with SummaryWriter(comment='Net1') as w:
    w.add_graph(model, (dummy_input, ))

Screenshots

Environment
tensorboard 1.13.1
tensorboardX 2.0
tensorflow 1.13.2
tensorflow-estimator 1.13.0
torch 1.4.0
torchvision 0.5.0

Python environment
Python 3.6.10 and conda 4.8.1

[lyq998]

I met the same trouble, and then I used the earlier version of pytorch and torchvision.
torch 1.3.0
torchvision 0.4.1
And I can see the graph in chrome now.
hope this helps

[kwhkim]

I am having the same problem,
with python 3.6.10, torch 1.4.0, torchvision 0.5.0, tensorboard 2.1.0, tensorboardX 2.0

[ZizhouJia]

The function scopeName return None in pytorch. It is likely pytorch remove the scopeName in pytorch==1.4

[hellwrld]

yes,with torch 1.3.0, torchvision 0.4.0,i see the graph

[b675987273]

The function scopeName return None in pytorch. It is likely pytorch remove the scopeName in pytorch==1.4

Why did they delete it?

[Jackintheboxes]

I met the same problem, with torch 1.5.0, torchvision 0.6.0.
I tried torch 1.3.0, torchvision 0.4.0 and it worked in Edge

[liuyangly1]

it works for torch 1.3.0 and torchvision 0.4.1.

[BiuBiu-pa]

Thank you all guys, I met the same situation,with torch 1.2.0 tensorboardX2.1 and I replaced the torch 1.2.0 with torch 1.3.0. It works!

[lanpa]

Hi, all. Since tensorboardX v2.1, the add_graph function is actually calling the graph generating function in pytorch. Please update to the latest pytorch and file an issue at pytorch/pytorch. Btw, the old add_graph() has been renamed to add_graph_deprecated().