Faych
Faych
I guess it caused by invalid file path, did ur dictionary contains "
Hi @ravenscroftj Thx for ur reply ! N sorry for my forgiveness of an image.  I 've do the instruction './codegen-serve -m ./models/codegen-2B-multi-ggml-4bit-quant.bin' and the image1 decribes what happening...
> This assert is triggering on my end - it looks like it may be expected to see an overwrite in the incoming stream? Or maybe we should just check...
I add a clean_filename function to remove the invalid character.🧹
thx, fix it now
Could u provide your torch version and ensure that match with you CUDA.
Sorry, my code is base on the SAM2 which is base on torch 2.3.1.
  Me too, trying to find another way out.
In the Traceback 7 and 8 we can see the error came from the transform, which came from: ```python self.transforms = torch.jit.script( nn.Sequential( Resize((self.resolution, self.resolution)), Normalize(self.mean, self.std), ) ) ```...
> Maybe the core code is from https://github.com/storyicon/comfyui_segment_anything?tab=readme-ov-file#comfyui-segment-anything and that's a project from half a year ago. many things have changed. not really, the backend is update to SAM2 while...