EinopsError raised

After uploading a picture, this error is raised:

File "C:\Users\zhang\anaconda3\lib\site-packages\einops\einops.py", line 418, in reduce
    raise EinopsError(message + '\n {}'.format(e))
einops.EinopsError:  Error while processing rearrange-reduction pattern "b n (h d) -> b h n d".
 Input tensor shape: torch.Size([1, 512]). Additional info: {'h': 8}.
 Expected 3 dimensions, got 2

This error also appears on google colab:

Saving Screenshot 2022-07-12 113227.png to Screenshot 2022-07-12 113227.png
---------------------------------------------------------------------------
EinopsError                               Traceback (most recent call last)
[/usr/local/lib/python3.7/dist-packages/einops/einops.py](https://localhost:8080/#) in reduce(tensor, pattern, reduction, **axes_lengths)
    409         recipe = _prepare_transformation_recipe(pattern, reduction, axes_lengths=hashable_axes_lengths)
--> 410         return _apply_recipe(recipe, tensor, reduction_type=reduction)
    411     except EinopsError as e:

17 frames
EinopsError: Expected 3 dimensions, got 2

During handling of the above exception, another exception occurred:

EinopsError                               Traceback (most recent call last)
[/usr/local/lib/python3.7/dist-packages/einops/einops.py](https://localhost:8080/#) in reduce(tensor, pattern, reduction, **axes_lengths)
    416             message += '\n Input is list. '
    417         message += 'Additional info: {}.'.format(axes_lengths)
--> 418         raise EinopsError(message + '\n {}'.format(e))
    419 
    420 

EinopsError:  Error while processing rearrange-reduction pattern "b n (h d) -> b h n d".
 Input tensor shape: torch.Size([1, 512]). Additional info: {'h': 8}.
 Expected 3 dimensions, got 2

[pkjq11]

In https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py The default value of parameter global_pool of VisionTransformer is “token”

def forward_head(self, x, pre_logits: bool = False):
        if self.global_pool:
            x = x[:, self.num_prefix_tokens:].mean(dim=1) if self.global_pool == 'avg' else x[:, 0]
        x = self.fc_norm(x)
        return x if pre_logits else self.head(x)

So we can add an argument (global_pool=“”) to construct CustomVisionTransformer in “hybrid.py”, like this

[lukas-blecher]

Will push the update to pypi when I get home. Until then you can call

pip install -U git+https://github.com/lukas-blecher/LaTeX-OCR.git

Edit: Done