Open glitchyg opened 1 year ago
I have been setting this up on my mac, and I wanted to give some details on issues I had and what I did to fix it:
Error: Error: buffer is not large enough. Must be 307200 bytes
Error: buffer is not large enough. Must be 307200 bytes
Solution: Caused by pytorch bug on M1 Mac, Upgraded my pytorch and vision pip install torch==1.13 torchvision==0.14.1
pip install torch==1.13 torchvision==0.14.1
This was caused by some code calling .cuda() clipcrop.py, line 78
clipcrop.py, line 78
I fixed it with this code:
device_name = "cuda" if torch.backends.mps.is_available(): device_name = "mps" device = torch.device(device_name) image_features.cpu().numpy() image_mean = torch.tensor( [0.48145466, 0.4578275, 0.40821073], device=device) image_std = torch.tensor( [0.26862954, 0.26130258, 0.27577711], device=device) images = [preprocess(im) for im in l] image_input = torch.tensor(np.stack(images), device=device)
After this i was able to get it running. Thanks for the awesome tool!
Thank for this - I think this has worked for me
I have been setting this up on my mac, and I wanted to give some details on issues I had and what I did to fix it:
Loading Yolo model on crop:
Error:
Error: buffer is not large enough. Must be 307200 bytes
Solution: Caused by pytorch bug on M1 Mac, Upgraded my pytorch and vision
pip install torch==1.13 torchvision==0.14.1
Torch not compiled with CUDA enabled
This was caused by some code calling .cuda()
clipcrop.py, line 78
I fixed it with this code:
After this i was able to get it running. Thanks for the awesome tool!