Failed to import diffusers.schedulers.scheduling_ddim because of the following error (look up to see its traceback): numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

[FurkanGozukara]

Describe the bug

Google Colab upgraded to Python 3.11 and now this basic code fails. I don't get why

Reproduction

import torch
from diffusers import StableDiffusionPipeline

# Path to your trained model
model_path = '/content/stable_diffusion_weights/ohwx/800'  # Update if needed

pipe = StableDiffusionPipeline.from_pretrained(
    "/content/stable_diffusion_weights/ohwx/800",
    torch_dtype=torch.float16,
        safety_checker=None,
)


 
prompt = "a photo of an astronaut riding a horse on mars"
pipe.enable_attention_slicing()
image = pipe(prompt).images[0]

Logs

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
/usr/local/lib/python3.11/dist-packages/diffusers/utils/import_utils.py in _get_module(self, module_name)
    717         try:
--> 718             return importlib.import_module("." + module_name, self.__name__)
    719         except Exception as e:

45 frames
numpy/random/mtrand.pyx in init numpy.random.mtrand()

ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

The above exception was the direct cause of the following exception:

RuntimeError                              Traceback (most recent call last)
RuntimeError: Failed to import diffusers.schedulers.scheduling_ddim because of the following error (look up to see its traceback):
numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

The above exception was the direct cause of the following exception:

RuntimeError                              Traceback (most recent call last)
RuntimeError: Failed to import diffusers.loaders.single_file because of the following error (look up to see its traceback):
Failed to import diffusers.schedulers.scheduling_ddim because of the following error (look up to see its traceback):
numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

The above exception was the direct cause of the following exception:

RuntimeError                              Traceback (most recent call last)
/usr/local/lib/python3.11/dist-packages/diffusers/utils/import_utils.py in _get_module(self, module_name)
    718             return importlib.import_module("." + module_name, self.__name__)
    719         except Exception as e:
--> 720             raise RuntimeError(
    721                 f"Failed to import {self.__name__}.{module_name} because of the following error (look up to see its"
    722                 f" traceback):\n{e}"

RuntimeError: Failed to import diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion because of the following error (look up to see its traceback):
Failed to import diffusers.loaders.single_file because of the following error (look up to see its traceback):
Failed to import diffusers.schedulers.scheduling_ddim because of the following error (look up to see its traceback):
numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

System Info

Copy-and-paste the text below in your GitHub issue and FILL OUT the two last points.

• diffusers version: 0.27.0
• Platform: Linux-6.1.85+-x86_64-with-glibc2.35
• Python version: 3.11.11
• PyTorch version (GPU?): 2.2.0+cu121 (True)
• Huggingface_hub version: 0.25.2
• Transformers version: 4.43.0
• Accelerate version: 1.5.2
• xFormers version: 0.0.24
• Using GPU in script?:
• Using distributed or parallel set-up in script?:

Who can help?

No response

[yiyixuxu]

is this specific to colab?

[FurkanGozukara]

is this specific to colab?

i only tested on colab so far could be. when it was python 3.10 it was working perfectly fine. they very recently migrated to python 3.11

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.

[MEI-JL]

Can confirm. I was trying to replicate the Diffusion policy on colab and I encountered the same issue. I am pretty new to colab environments, though I came across another issue and its comment:
ostris/ai-toolkit#267 (comment)

It is a problem with numpy version and can be solved by downgrading and a forced runtime restart. At first I tried to downgrade numpy alone without runtime restart and it will not work. Following all steps in the comment solved the import problem for me. (though I have not tested other functionalities, would update after I do so)

edit: my network is training so I assume that there are no other problems

[FurkanGozukara]

Can confirm. I was trying to replicate the Diffusion policy on colab and I encountered the same issue. I am pretty new to colab environments, though I came across another issue and its comment: ostris/ai-toolkit#267 (comment)

It is a problem with numpy version and can be solved by downgrading and a forced runtime restart. At first I tried to downgrade numpy alone without runtime restart and it will not work. Following all steps in the comment solved the import problem for me. (though I have not tested other functionalities, would update after I do so)

wow nice solution i will test now

[FurkanGozukara]

Can confirm. I was trying to replicate the Diffusion policy on colab and I encountered the same issue. I am pretty new to colab environments, though I came across another issue and its comment: ostris/ai-toolkit#267 (comment)

It is a problem with numpy version and can be solved by downgrading and a forced runtime restart. At first I tried to downgrade numpy alone without runtime restart and it will not work. Following all steps in the comment solved the import problem for me. (though I have not tested other functionalities, would update after I do so)

this worked ty so much

[yiyixuxu]

closing as it is resolved!
thanks @FurkanGozukara @MEI-JL

[FeiCuiLeng]

I encountered the same error in my project, but not in Colab. I resolved it by downgrading numpy to version 1.26.3. I also noticed that the issue might be caused by an error during the installation of the numpy package, which could lead to version mismatches between numpy and numpy-base. This inconsistency might also cause the error. You can try uninstalling numpy multiple times, which might help resolve the issue.

Error message:
Failed to import diffusers.schedulers.scheduling_dpmsolver_multistep because of the following error (look up to see its traceback):
numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject.

Version information for the fix:

• Ubuntu 22.04
• Python 3.10
• numpy: 1.26.3
• diffusers: 0.33.1