search

ssube / onnx-web

web UI for GPU-accelerated ONNX pipelines like Stable Diffusion, even on Windows and AMD
https://discord.gg/7CdQmutGuw
MIT License
204 stars 26 forks source link

RunwayML took down their Huggingface resources #468

Open kynalvarus opened 1 month ago

kynalvarus commented 1 month ago

See https://huggingface.co/runwayml

Note: We are no longer maintaining a HuggingFace organization. You can explore more recent open-source image generation models here: https://huggingface.co/models

Error during setup_amd.bat: [2024-09-15 12:32:25,627] INFO: MainProcess MainThread onnx_web.convert.client.huggingface: downloading model from Huggingface Hub: runwayml/stable-diffusion-v1-5 Couldn't connect to the Hub: 404 Client Error. (Request ID: Root=1-66e70c19-02e8fe631dc7766b0465de4f;6f827bc5-a954-4815-bf39-9fd342f610dd). ... [2024-09-15 12:32:25,879] INFO: MainProcess MainThread onnx_web.convert.client.huggingface: downloading model from Huggingface Hub: runwayml/stable-diffusion-inpainting Couldn't connect to the Hub: 404 Client Error. (Request ID: Root=1-66e70c19-001a3dd27ff036553cce6603;84e7f6a6-2c82-4ebb-aa1c-0df2eb4b2d35)

I'm looking for an alternate source for these, if you haven't already.

kynalvarus commented 1 month ago

Patching based on https://www.reddit.com/r/StableDiffusion/comments/1f4epto/runway_took_down_15_and_15_inpainting/

kynalvarus commented 1 month ago

stable-diffusion-v1-5 is fine from the new repo, but inpainting won't convert.

[2024-09-15 14:54:55,008] INFO: MainProcess MainThread onnx_web.convert.client.huggingface: downloading model from Huggingface Hub: benjamin-paine/stable-diffusion-v1-5-inpainting
Keyword arguments {'extract_ema': True} are not expected by StableDiffusionPipeline and will be ignored.
Loading pipeline components...:  14%|█████████████▍                                                                                | 1/7 [00:00<00:00, 249.44it/s] 
[2024-09-15 14:54:55,401] ERROR: MainProcess MainThread __main__: error converting diffusion model stable-diffusion-onnx-v1-inpainting
Traceback (most recent call last):
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_web\convert\__main__.py", line 458, in convert_models
    convert_model_diffusion(conversion, model)
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_web\convert\__main__.py", line 278, in convert_model_diffusion
    converted, dest = converter(
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_env\lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_web\convert\diffusion\diffusion.py", line 808, in convert_diffusion_diffusers_optimum
    pipeline = pipe_class.from_pretrained(cache_path, **pipe_args)
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_env\lib\site-packages\diffusers\pipelines\pipeline_utils.py", line 1265, in from_pretrained      
    loaded_sub_model = load_sub_model(
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_env\lib\site-packages\diffusers\pipelines\pipeline_utils.py", line 520, in load_sub_model        
    loaded_sub_model = load_method(os.path.join(cached_folder, name), **loading_kwargs)
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_env\lib\site-packages\transformers\modeling_utils.py", line 3362, in from_pretrained
    if metadata.get("format") == "pt":
AttributeError: 'NoneType' object has no attribute 'get'

Not sure what's wrong, but it looks like a missing file or something similar - f.metadata is NoneType when it should not be after safe_load.

I'll throw my changes up as a draft PR.

kynalvarus commented 1 month ago

Draft PR: https://github.com/ssube/onnx-web/pull/469

kynalvarus commented 1 month ago

Switched to the botp repo on huggingface, and that seemed to succeed through the ONNX conversion. However, test_diffusers still fails:

python .\scripts\test-diffusers.py
generating test image...
Loading pipeline components...:   0%|                                                                                                       | 0/5 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\scripts\test-diffusers.py", line 20, in <module>
    pipe = OnnxStableDiffusionPipeline.from_pretrained(model, provider='DmlExecutionProvider', safety_checker=None)
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_env\lib\site-packages\diffusers\pipelines\pipeline_utils.py", line 1265, in from_pretrained      
    loaded_sub_model = load_sub_model(
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_env\lib\site-packages\diffusers\pipelines\pipeline_utils.py", line 520, in load_sub_model        
    loaded_sub_model = load_method(os.path.join(cached_folder, name), **loading_kwargs)
  File "C:\Users\Ryan\Documents\Projects\imggen\onnx-web\api\onnx_env\lib\site-packages\transformers\modeling_utils.py", line 3194, in from_pretrained
    raise EnvironmentError(
OSError: Error no file named pytorch_model.bin, tf_model.h5, model.ckpt.index or flax_model.msgpack found in directory ..\models\stable-diffusion-onnx-v1-5\text_encoder.