Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

size mismatch when run "python i3d_tf_to_pt.py --rgb" #20

Open
liuyuemaicha opened this issue Nov 2, 2018 · 5 comments
Open

size mismatch when run "python i3d_tf_to_pt.py --rgb" #20

liuyuemaicha opened this issue Nov 2, 2018 · 5 comments

Comments

@liuyuemaicha
Copy link

Hi,
Thank you for your work, firstly.
I want to transfer the pre-training parameters in Tensorflow to PyTorch. But when I run "python i3d_tf_to_pt.py --rgb", I have the bugs as follows:
Additionally, I want to know, the pre-training parameters (model/model_rgb.pth) in PyTorch you provide if as same as the Tensorflow ? Thank you very much !


2018-11-02 16:23:32.948892: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Traceback (most recent call last):
  File "i3d_tf_to_pt.py", line 177, in <module>
    modality='rgb')
  File "i3d_tf_to_pt.py", line 73, in transfer_weights
    i3nception_pt.load_tf_weights(sess)
  File "/Users/cuixiankun01/project/baidu/kinetics_i3d_pytorch-master/src/i3dpt.py", line 310, in load_tf_weights
    self.load_state_dict(state_dict)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/torch/nn/modules/module.py", line 719, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for I3D:
        size mismatch for conv3d_1a_7x7.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_1a_7x7.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_1a_7x7.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2b_1x1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2b_1x1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2b_1x1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2c_3x3.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for conv3d_2c_3x3.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for conv3d_2c_3x3.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3b.branch_0.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3b.branch_0.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3b.branch_0.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3b.branch_1.0.batch3d.running_var: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3b.branch_1.0.batch3d.bias: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3b.branch_1.0.batch3d.running_mean: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3b.branch_1.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3b.branch_1.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3b.branch_1.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3b.branch_2.0.batch3d.running_var: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_3b.branch_2.0.batch3d.bias: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_3b.branch_2.0.batch3d.running_mean: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_3b.branch_2.1.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_2.1.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_2.1.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_3.1.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_3.1.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_3.1.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.1.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3c.branch_1.1.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3c.branch_1.1.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3c.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_2.1.batch3d.running_var: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3c.branch_2.1.batch3d.bias: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3c.branch_2.1.batch3d.running_mean: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3c.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3c.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3c.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4b.branch_0.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_4b.branch_0.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_4b.branch_0.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_4b.branch_1.0.batch3d.running_var: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_4b.branch_1.0.batch3d.bias: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_4b.branch_1.0.batch3d.running_mean: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_4b.branch_1.1.batch3d.running_var: copying a param of torch.Size([208]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 208]) in current model.
        size mismatch for mixed_4b.branch_1.1.batch3d.bias: copying a param of torch.Size([208]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 208]) in current model.
        size mismatch for mixed_4b.branch_1.1.batch3d.running_mean: copying a param of torch.Size([208]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 208]) in current model.
        size mismatch for mixed_4b.branch_2.0.batch3d.running_var: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_4b.branch_2.0.batch3d.bias: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_4b.branch_2.0.batch3d.running_mean: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_4b.branch_2.1.batch3d.running_var: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_4b.branch_2.1.batch3d.bias: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_4b.branch_2.1.batch3d.running_mean: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_4b.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4b.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4b.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_0.batch3d.running_var: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4c.branch_0.batch3d.bias: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4c.branch_0.batch3d.running_mean: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4c.branch_1.0.batch3d.running_var: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4c.branch_1.0.batch3d.bias: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4c.branch_1.0.batch3d.running_mean: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4c.branch_1.1.batch3d.running_var: copying a param of torch.Size([224]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 224]) in current model.
        size mismatch for mixed_4c.branch_1.1.batch3d.bias: copying a param of torch.Size([224]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 224]) in current model.
        size mismatch for mixed_4c.branch_1.1.batch3d.running_mean: copying a param of torch.Size([224]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 224]) in current model.
        size mismatch for mixed_4c.branch_2.0.batch3d.running_var: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4c.branch_2.0.batch3d.bias: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4c.branch_2.0.batch3d.running_mean: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4c.branch_2.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_2.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_2.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.1.batch3d.running_var: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4d.branch_1.1.batch3d.bias: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4d.branch_1.1.batch3d.running_mean: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4d.branch_2.0.batch3d.running_var: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4d.branch_2.0.batch3d.bias: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4d.branch_2.0.batch3d.running_mean: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4d.branch_2.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_2.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_2.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_0.batch3d.running_var: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4e.branch_0.batch3d.bias: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4e.branch_0.batch3d.running_mean: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4e.branch_1.0.batch3d.running_var: copying a param of torch.Size([144]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 144]) in current model.
        size mismatch for mixed_4e.branch_1.0.batch3d.bias: copying a param of torch.Size([144]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 144]) in current model.
        size mismatch for mixed_4e.branch_1.0.batch3d.running_mean: copying a param of torch.Size([144]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 144]) in current model.
        size mismatch for mixed_4e.branch_1.1.batch3d.running_var: copying a param of torch.Size([288]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 288]) in current model.
        size mismatch for mixed_4e.branch_1.1.batch3d.bias: copying a param of torch.Size([288]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 288]) in current model.
        size mismatch for mixed_4e.branch_1.1.batch3d.running_mean: copying a param of torch.Size([288]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 288]) in current model.
        size mismatch for mixed_4e.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4e.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4e.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4e.branch_2.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_2.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_2.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4f.branch_0.batch3d.running_var: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4f.branch_0.batch3d.bias: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4f.branch_0.batch3d.running_mean: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4f.branch_1.0.batch3d.running_var: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4f.branch_1.0.batch3d.bias: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4f.branch_1.0.batch3d.running_mean: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4f.branch_1.1.batch3d.running_var: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_4f.branch_1.1.batch3d.bias: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_4f.branch_1.1.batch3d.running_mean: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_4f.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4f.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4f.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4f.branch_2.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_2.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_2.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_3.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_3.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_3.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_0.batch3d.running_var: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_5b.branch_0.batch3d.bias: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_5b.branch_0.batch3d.running_mean: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_5b.branch_1.0.batch3d.running_var: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_5b.branch_1.0.batch3d.bias: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_5b.branch_1.0.batch3d.running_mean: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_5b.branch_1.1.batch3d.running_var: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_5b.branch_1.1.batch3d.bias: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_5b.branch_1.1.batch3d.running_mean: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_5b.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_5b.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_5b.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_5b.branch_2.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_2.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_2.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_3.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_3.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_3.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_0.batch3d.running_var: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_0.batch3d.bias: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_0.batch3d.running_mean: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_1.0.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_5c.branch_1.0.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_5c.branch_1.0.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_5c.branch_1.1.batch3d.running_var: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_1.1.batch3d.bias: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_1.1.batch3d.running_mean: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_2.0.batch3d.running_var: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_5c.branch_2.0.batch3d.bias: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_5c.branch_2.0.batch3d.running_mean: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_5c.branch_2.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_2.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_2.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_3.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_3.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_3.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
@windIsMyHeart
Copy link

Hi,
Thank you for your work, firstly.
I want to transfer the pre-training parameters in Tensorflow to PyTorch. But when I run "python i3d_tf_to_pt.py --rgb", I have the bugs as follows:
Additionally, I want to know, the pre-training parameters (model/model_rgb.pth) in PyTorch you provide if as same as the Tensorflow ? Thank you very much !


2018-11-02 16:23:32.948892: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Traceback (most recent call last):
  File "i3d_tf_to_pt.py", line 177, in <module>
    modality='rgb')
  File "i3d_tf_to_pt.py", line 73, in transfer_weights
    i3nception_pt.load_tf_weights(sess)
  File "/Users/cuixiankun01/project/baidu/kinetics_i3d_pytorch-master/src/i3dpt.py", line 310, in load_tf_weights
    self.load_state_dict(state_dict)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/torch/nn/modules/module.py", line 719, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for I3D:
        size mismatch for conv3d_1a_7x7.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_1a_7x7.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_1a_7x7.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2b_1x1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2b_1x1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2b_1x1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for conv3d_2c_3x3.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for conv3d_2c_3x3.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for conv3d_2c_3x3.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3b.branch_0.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3b.branch_0.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3b.branch_0.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3b.branch_1.0.batch3d.running_var: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3b.branch_1.0.batch3d.bias: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3b.branch_1.0.batch3d.running_mean: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3b.branch_1.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3b.branch_1.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3b.branch_1.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3b.branch_2.0.batch3d.running_var: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_3b.branch_2.0.batch3d.bias: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_3b.branch_2.0.batch3d.running_mean: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_3b.branch_2.1.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_2.1.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_2.1.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_3.1.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_3.1.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3b.branch_3.1.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_3c.branch_1.1.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3c.branch_1.1.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3c.branch_1.1.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_3c.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_3c.branch_2.1.batch3d.running_var: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3c.branch_2.1.batch3d.bias: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3c.branch_2.1.batch3d.running_mean: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_3c.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3c.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_3c.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4b.branch_0.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_4b.branch_0.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_4b.branch_0.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_4b.branch_1.0.batch3d.running_var: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_4b.branch_1.0.batch3d.bias: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_4b.branch_1.0.batch3d.running_mean: copying a param of torch.Size([96]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 96]) in current model.
        size mismatch for mixed_4b.branch_1.1.batch3d.running_var: copying a param of torch.Size([208]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 208]) in current model.
        size mismatch for mixed_4b.branch_1.1.batch3d.bias: copying a param of torch.Size([208]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 208]) in current model.
        size mismatch for mixed_4b.branch_1.1.batch3d.running_mean: copying a param of torch.Size([208]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 208]) in current model.
        size mismatch for mixed_4b.branch_2.0.batch3d.running_var: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_4b.branch_2.0.batch3d.bias: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_4b.branch_2.0.batch3d.running_mean: copying a param of torch.Size([16]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 16]) in current model.
        size mismatch for mixed_4b.branch_2.1.batch3d.running_var: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_4b.branch_2.1.batch3d.bias: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_4b.branch_2.1.batch3d.running_mean: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_4b.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4b.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4b.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_0.batch3d.running_var: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4c.branch_0.batch3d.bias: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4c.branch_0.batch3d.running_mean: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4c.branch_1.0.batch3d.running_var: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4c.branch_1.0.batch3d.bias: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4c.branch_1.0.batch3d.running_mean: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4c.branch_1.1.batch3d.running_var: copying a param of torch.Size([224]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 224]) in current model.
        size mismatch for mixed_4c.branch_1.1.batch3d.bias: copying a param of torch.Size([224]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 224]) in current model.
        size mismatch for mixed_4c.branch_1.1.batch3d.running_mean: copying a param of torch.Size([224]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 224]) in current model.
        size mismatch for mixed_4c.branch_2.0.batch3d.running_var: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4c.branch_2.0.batch3d.bias: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4c.branch_2.0.batch3d.running_mean: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4c.branch_2.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_2.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_2.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4c.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.0.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.0.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.0.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4d.branch_1.1.batch3d.running_var: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4d.branch_1.1.batch3d.bias: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4d.branch_1.1.batch3d.running_mean: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4d.branch_2.0.batch3d.running_var: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4d.branch_2.0.batch3d.bias: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4d.branch_2.0.batch3d.running_mean: copying a param of torch.Size([24]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 24]) in current model.
        size mismatch for mixed_4d.branch_2.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_2.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_2.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4d.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_0.batch3d.running_var: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4e.branch_0.batch3d.bias: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4e.branch_0.batch3d.running_mean: copying a param of torch.Size([112]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 112]) in current model.
        size mismatch for mixed_4e.branch_1.0.batch3d.running_var: copying a param of torch.Size([144]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 144]) in current model.
        size mismatch for mixed_4e.branch_1.0.batch3d.bias: copying a param of torch.Size([144]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 144]) in current model.
        size mismatch for mixed_4e.branch_1.0.batch3d.running_mean: copying a param of torch.Size([144]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 144]) in current model.
        size mismatch for mixed_4e.branch_1.1.batch3d.running_var: copying a param of torch.Size([288]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 288]) in current model.
        size mismatch for mixed_4e.branch_1.1.batch3d.bias: copying a param of torch.Size([288]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 288]) in current model.
        size mismatch for mixed_4e.branch_1.1.batch3d.running_mean: copying a param of torch.Size([288]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 288]) in current model.
        size mismatch for mixed_4e.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4e.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4e.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4e.branch_2.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_2.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_2.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_3.1.batch3d.running_var: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_3.1.batch3d.bias: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4e.branch_3.1.batch3d.running_mean: copying a param of torch.Size([64]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 64]) in current model.
        size mismatch for mixed_4f.branch_0.batch3d.running_var: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4f.branch_0.batch3d.bias: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4f.branch_0.batch3d.running_mean: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_4f.branch_1.0.batch3d.running_var: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4f.branch_1.0.batch3d.bias: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4f.branch_1.0.batch3d.running_mean: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_4f.branch_1.1.batch3d.running_var: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_4f.branch_1.1.batch3d.bias: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_4f.branch_1.1.batch3d.running_mean: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_4f.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4f.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4f.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_4f.branch_2.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_2.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_2.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_3.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_3.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_4f.branch_3.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_0.batch3d.running_var: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_5b.branch_0.batch3d.bias: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_5b.branch_0.batch3d.running_mean: copying a param of torch.Size([256]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 256]) in current model.
        size mismatch for mixed_5b.branch_1.0.batch3d.running_var: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_5b.branch_1.0.batch3d.bias: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_5b.branch_1.0.batch3d.running_mean: copying a param of torch.Size([160]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 160]) in current model.
        size mismatch for mixed_5b.branch_1.1.batch3d.running_var: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_5b.branch_1.1.batch3d.bias: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_5b.branch_1.1.batch3d.running_mean: copying a param of torch.Size([320]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 320]) in current model.
        size mismatch for mixed_5b.branch_2.0.batch3d.running_var: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_5b.branch_2.0.batch3d.bias: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_5b.branch_2.0.batch3d.running_mean: copying a param of torch.Size([32]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 32]) in current model.
        size mismatch for mixed_5b.branch_2.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_2.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_2.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_3.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_3.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5b.branch_3.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_0.batch3d.running_var: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_0.batch3d.bias: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_0.batch3d.running_mean: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_1.0.batch3d.running_var: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_5c.branch_1.0.batch3d.bias: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_5c.branch_1.0.batch3d.running_mean: copying a param of torch.Size([192]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 192]) in current model.
        size mismatch for mixed_5c.branch_1.1.batch3d.running_var: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_1.1.batch3d.bias: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_1.1.batch3d.running_mean: copying a param of torch.Size([384]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 384]) in current model.
        size mismatch for mixed_5c.branch_2.0.batch3d.running_var: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_5c.branch_2.0.batch3d.bias: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_5c.branch_2.0.batch3d.running_mean: copying a param of torch.Size([48]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 48]) in current model.
        size mismatch for mixed_5c.branch_2.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_2.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_2.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_3.1.batch3d.running_var: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_3.1.batch3d.bias: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.
        size mismatch for mixed_5c.branch_3.1.batch3d.running_mean: copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128]) in current model.

你解决这个问题了吗

@hassony2
Copy link
Owner

hassony2 commented Nov 6, 2018 •
edited
Loading

Hi !

If you look at the error the size mismatches are between tensors which have the same number of elements, also they only happen on the batchnorm parameters.
Like "copying a param of torch.Size([128]) from checkpoint, where the shape is torch.Size([1, 1, 1, 1, 128])". My guess is that it might be a pytorch version issue. Which one are you using ?

An alternate solution might be to add the extra dimensions (using something like view(1, 1, 1, 1, -1) on each of the weights obtained from tensorflow probably somewhere here

Right now is pretty busy and I won't be able to look more into depth into this (meaning actually running code, experimenting with pytorch versions...) before the next 10 days :)

Let me know if you could solve this :)

Best !

Yana

liuyuemaicha added a commit to liuyuemaicha/kinetics_i3d_pytorch that referenced this issue Nov 7, 2018
@liuyuemaicha
Update i3dpt.py
71df19c 
fix the bug "size mismatch",bug description:  hassony2#20
@liuyuemaicha liuyuemaicha mentioned this issue Nov 7, 2018
Open
@liuyuemaicha
Copy link
Author

liuyuemaicha commented Nov 7, 2018 •
edited
Loading

Hi,
Thanks for your advise.
I've found the problem. the shape of parameters from Tensorflow is [1, 1, 1, 1, n], mismatch the size from PyTorch ([n]).
I've fixed the bug and make a pull request.
Thanks for your help and great work !
谢谢😁

$ python i3d_tf_to_pt.py --rgb
2018-11-07 15:09:42.636465: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA

max diff : 0.0115141868591, mean diff : 2.91702690447e-05
mean val: tf 0.00250000040978 pt 0.00250000040978
max vals: tf 0.943557560444 pt 0.943576276302
max relative diff: tf 0.21941703558 pt 0.281093806028

max diff : 0.00800371170044, mean diff : 4.55450135632e-05
mean val: tf 0.00249999971129 pt 0.00249999994412
max vals: tf 0.602078437805 pt 0.610082149506
max relative diff: tf 0.148675397038 pt 0.174640089273

max diff : 0.0120238959789, mean diff : 4.93937077408e-05
mean val: tf 0.00249999994412 pt 0.00249999994412
max vals: tf 0.355097353458 pt 0.367121249437
max relative diff: tf 0.120408758521 pt 0.136891722679

@hassony2
Copy link
Owner

hassony2 commented Nov 8, 2018

Thank you for the pull request !

Before accepting it, I would like to look at what causes the problem (I still think it might be pytorch or tensorflow-version specific, so best before I merge would be to understand why it happened to you and not to me ! :) )

Best !

Yana

@liuyuemaicha
Copy link
Author

Yes, you reminded me ! I updated my Pytorch from 0.3 to 0.4.1 recently, and yesterday I found the similar problem in another program.
Thank you for your help even you are very busy now, I hope you will have a great reward !😊

Best ！
Kun

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@hassony2 @liuyuemaicha @windIsMyHeart