over a year late but we converted SigLIP 2 to Core ML and sorted 7,349 pet photos into 37 breeds on a M5 Coreml model vs. base PyTorch model, same accuracy: • 36 s vs 102 s • 5 ms per photo vs 14 ms • 262 MB peak RAM vs 4.2 GB • 715 MB on disk vs 1.5 GB model:

Name: over a year late but we converted SigLIP 2 to Core ML and sorted 7,349 pet photos into 37 breeds on a M5 Coreml model vs. base PyTorch model, same accuracy: • 36 s vs 102 s • 5 ms per photo vs 14 ms • 262 MB peak RAM vs 4.2 GB • 715 MB on disk vs 1.5 GB model:
Channel: Alex
Description: over a year late but we converted SigLIP 2 to Core ML and sorted 7,349 pet photos into 37 breeds on a M5 Coreml model vs. base PyTorch model, same accuracy: • 36 s vs 102 s • 5 ms per photo vs 14 ms • 262 MB peak RAM vs 4.2 GB • 715 MB on disk vs 1.5 GB model:

@Alex

1 grabs

Download

over a year late but we converted SigLIP 2 to Core ML and sorted 7,349 pet photos into 37 breeds on a M5 Coreml model vs. base PyTorch model, same accuracy: • 36 s vs 102 s • 5 ms per photo vs 14 ms • 262 MB peak RAM vs 4.2 GB • 715 MB on disk vs 1.5 GB model:

No comments yet