RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

[dmavromatis]

Describe the bug When I try to generate I get the following error:

Error completing request Arguments: ('Test', '', 'None', 'None', 20, 0, False, False, 1, 1, 7, -1.0, -1.0, 0, 0, 0, False, 512, 512, False, 0.7, 0, 0, 0, False, False, None, '', 1, '', 0, '', True, False, False) {}

RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

File "/usr/local/lib/python3.10/site-packages/torch/nn/functional.py", line 2503, in layer_norm return torch.layer_norm(input, normalized_shape, weight, bias, eps, torch.backends.cudnn.enabled) RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

Mac M1

[jabelardo]

I have the same error after the last git pull

Windows 10

[dmavromatis]

Makes sense -- because I had generated 1 successfully, this morning -- then started it back up and got that error.

[cobryan05]

modules\textual_inversion\image_embedding.py, insert_image_data_embed(),

change

    data_np_low.resize(next_size)
    data_np_low = data_np_low.reshape((h, -1, d))

    data_np_high.resize(next_size)
    data_np_high = data_np_high.reshape((h, -1, d))

to

    np.resize(data_np_low,next_size)
    data_np_low = data_np_low.reshape((h, -1, d))

    np.resize(data_np_high, next_size)
    data_np_high = data_np_high.reshape((h, -1, d))

[soulayman-al-abdallah]

The solution is mentioned here: https://github.com/Birch-san/stable-diffusion#patch

all what you have to do is going the file mentioned by the error, i.e. from cli: open /usr/local/lib/python3.10/site-packages/torch/nn/functional.py or in my ios case in a conda env: open /opt/homebrew/Caskroom/miniforge/base/envs/web-ui/lib/python3.10/site-packages/torch/nn/functional.py

and when it opens go to the mentioned line 2503 to change input to input..contiguous()

save, rerun the app and it worked 👍

the problem appeard on my mac M1 but not on colab

[dgolant]

Happened to me today on m2 macbook air.

[lzhbhlrPython]

Happened to me today on m1pro MacBook Pro.

[ersingencturk]

The solution is mentioned here: https://github.com/Birch-san/stable-diffusion#patch

all what you have to do is going the file mentioned by the error, i.e. from cli: open /usr/local/lib/python3.10/site-packages/torch/nn/functional.py or in my ios case in a conda env: open /opt/homebrew/Caskroom/miniforge/base/envs/web-ui/lib/python3.10/site-packages/torch/nn/functional.py

and when it opens go to the mentioned line 2503 to change input to input..contiguous()

save, rerun the app and it worked 👍

the problem appeard on my mac M1 but not on colab

this worked for me, however looks like it has been fixed already in torch 1.13.1