discordinated

Results 4 comments of discordinated

Seems to be this part of comfy/model_management.py ``` def should_use_bf16(device=None): if is_intel_xpu(): return True if device is None: device = torch.device("cuda") props = torch.cuda.get_device_properties(device) if props.major >= 8: return True...

That got past that part of the problem I'm now getting this even running with --force-fp32 Pipelines loaded with `dtype=torch.float16` cannot run with `cpu` device. It is not recommended to...

I'm on Mac M1 Max btw. Maybe this requires CUDA?

Same on beta 15.4... across invoke + comfyui