failed assertion `Placeholder shape mismatches (expected 1 vs got tensorData with 896) at dimIdx = 0' when recognizing text

[gaowq2017]

The detial as follow:
OS: Max os, chip: M1
ErrorLog:
Loaded detection model vikp/surya_det3 on device mps with dtype torch.float16
Loaded recognition model vikp/surya_rec2 on device mps with dtype torch.float16
Detecting bboxes: 100%|██████████| 1/1 [00:00<00:00, 1.55it/s]
Recognizing Text: 0%| | 0/1 [00:00<?, ?it/s]/AppleInternal/Library/BuildRoots/d9889869-120b-11ee-b796-7a03568b17ac/Library/Caches/com.apple.xbs/Sources/MetalPerformanceShadersGraph/mpsgraph/MetalPerformanceShadersGraph/Core/Files/MPSGraph.mm:678: failed assertion `Placeholder shape mismatches (expected 1 vs got tensorData with 896) at dimIdx = 0'
/Users/theoyu/miniforge3/envs/py310/lib/python3.10/multiprocessing/resource_tracker.py:224: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
warnings.warn('resource_tracker: There appear to be %d '
Thank you!

[VikParuchuri]

This is an MPS bug, should be fixed with the newest version of surya, make sure you do pip install -U

[gaowq2017]

This is an MPS bug, should be fixed with the newest version of surya, make sure you do pip install -U

@VikParuchuri OK, thank you very much!

[olofssonmattias]

Hello I'm getting this error but pip install -U does nothing unfortuantely for me.

Kind Regards

Matt

[jwuphysics]

Unfortunately I am still seeing this issue with a fresh install in a new virtual environment. Same error with Metal as above:

Loaded layout model datalab-to/surya_layout on device mps with dtype torch.float16
Loaded texify model datalab-to/texify on device mps with dtype torch.float16
Loaded recognition model vikp/surya_rec2 on device mps with dtype torch.float16
Loaded table recognition model datalab-to/surya_tablerec on device mps with dtype torch.float16
Loaded detection model vikp/surya_det3 on device mps with dtype torch.float16
Recognizing layout:   0%|                                                                                                                                        | 0/6 [00:00<?, ?it/s]/AppleInternal/Library/BuildRoots/b11baf73-9ee0-11ef-b7b4-7aebe1f78c73/Library/Caches/com.apple.xbs/Sources/MetalPerformanceShadersGraph/mpsgraph/MetalPerformanceShadersGraph/Core/Files/MPSGraph.mm:971: failed assertion `Placeholder shape mismatches (expected 1 vs got tensorData with 3456) at dimIdx = 0'
fish: Job 1, 'marker_single ./arxiv-papers/22…' terminated by signal SIGABRT (Abort)
/Users/john/miniforge3/envs/jwst-dois/lib/python3.11/multiprocessing/resource_tracker.py:254: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
  warnings.warn('resource_tracker: There appear to be %d '

[kbruegge]

I'm having the same issue for my input. Bounding boxes seem to be fine from visual inspection.

[kbruegge]

Is there a way to force execution on the CPU instead of the MPS device?

[VikParuchuri]

This was an issue with pytorch 2.6 - should be fixed with surya-ocr 0.10.2