No such file or directory: 'diffusers-cache/models--CompVis--stable-diffusion-safety-checker/snapshots/cb41f3a270d63d454d385fc2e4f571c487c253c5/config.json' about cog-stable-diffusion HOT 11 CLOSED

think it is the issue is from cache_dir, here is an alternative way to load and save the model https://github.com/chenxwh/cog-stable-diffusion/tree/tmp_fix_cachedir

from cog-stable-diffusion.

looks like this issue comes from huggingface_hub >=0.13, which was released yesterday. Fun! specifying huggingface-hub==0.12.1 should also fix it.

from cog-stable-diffusion.

I tried disabling safety checker code but still seeing a similar error:

Downloading (…)ain/model_index.json: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 539/539 [00:00<00:00, 543kB/s]
Traceback (most recent call last):
  File "/src/script/download-weights", line 24, in <module>
    pipe = StableDiffusionPipeline.from_pretrained(
  File "/root/.pyenv/versions/3.10.10/lib/python3.10/site-packages/diffusers/pipeline_utils.py", line 459, in from_pretrained
    config_dict = cls.load_config(
  File "/root/.pyenv/versions/3.10.10/lib/python3.10/site-packages/diffusers/configuration_utils.py", line 380, in load_config
    config_dict = cls._dict_from_json_file(config_file)
  File "/root/.pyenv/versions/3.10.10/lib/python3.10/site-packages/diffusers/configuration_utils.py", line 479, in _dict_from_json_file
    with open(json_file, "r", encoding="utf-8") as reader:
FileNotFoundError: [Errno 2] No such file or directory: 'diffusers-cache/models--stabilityai--stable-diffusion-2-1/snapshots/36a01dc742066de2e8c91e7cf0b8f6b53ef53da1/model_index.json'
ⅹ exit status 1

from cog-stable-diffusion.

I see the cached files all greyed out, and some indication that there are symlinks... ?

from cog-stable-diffusion.

Getting the same error on a Lambda Labs machine. Looks like something is up with that config.json file:

from cog-stable-diffusion.

weird! I just ran it from main without issue. perhaps we can figure it out when pairing tomorrow.

the symlink from snapshots to blobs is expected, though perhaps a fun surprise.

from cog-stable-diffusion.

Hi,
I just recently started getting this kind of error as well. I always remove the safety checker as I am pulling custom SD models, looks like I am unable to download any SD weights from HuggingFace now.

Last time I successfully downloaded the same weights were a couple of days ago. Could it be something in HF that has changed?

Tried local Windows machine and Lambda labs.

from cog-stable-diffusion.

Looks like it's fixed in 0.13.2: https://github.com/huggingface/huggingface_hub/releases/tag/v0.13.2

from cog-stable-diffusion.

specifying huggingface-hub==0.12.1 should also fix it.

How do we do that? huggingface-hub is not specified in cog.yaml. It's a transitive dependency, presumably of diffusers. Does adding it to build.python_packages in cog.yaml take precedence?

from cog-stable-diffusion.

How do we do that? huggingface-hub is not specified in cog.yaml. It's a transitive dependency, presumably of diffusers. Does adding it to build.python_packages in cog.yaml take precedence?

I believe so if you specify huggingface-hub after diffuser in cog.yaml

from cog-stable-diffusion.

☝🏼 Yep that worked. 4ef4734

from cog-stable-diffusion.