Closed raisindetre closed 10 months ago
same error RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.
I've got exactly the same error on a Apple M1 Max (32GB ram) macbook pro.
A current workaround that has worked for me - need to do more testing to see if it's consistent - is to add the command-line flag "--no-half" to your launch settings. This has provided reasonable results and suppressed errors in the console. Interesting though, that the issue only occurs for these models.
the same error even with --no half on, a few days ago it was working
--no-half didn't help my setup:
M1 16gb
version: [v1.6.0] • python: 3.10.9 • torch: 2.0.1 • xformers: N/A • gradio: 3.41.2 • checkpoint: [9aba26abdf]
ControlNet v1.1.410
Is there an existing issue for this?
What happened?
Installed SD1.5 ip-adapter_* models from https://huggingface.co/lllyasviel/sd_control_collection/tree/main to v1.60 Automatic1111 using CN v1.1.409 and restarted. Using IP-Adapter within Control Net generates random images and an "Attempting to deserialize object on a CUDA device" runtime error. Other models tested (Reference, OpenPose) working as expected.
Steps to reproduce the problem
What should have happened?
New image generated similar to image provided to Control Net.
Commit where the problem happens
webui: 5ef669de080814067961f28357256e8fe27544f4 controlnet: b15636ed35eff934af69985bcdfbc407cfedfe7d
What browsers do you use to access the UI ?
Google Chrome
Command Line Arguments
List of enabled extensions
Console logs
Additional information
No response