Thanks for your PR. This would likely break the CoreML export due to the interpolation with dynamic size. I'd appreciate it if you could split the usage of SamOnnxModel from the current CoreML export, like adding a SamCoreMLModel separately for the CoreML export using the current sam/repvit_sam/utils/onnx.py and keeping the SamOnnxModel the same as your PR.
Thanks for your PR. This would likely break the CoreML export due to the interpolation with dynamic size. I'd appreciate it if you could split the usage of SamOnnxModel from the current CoreML export, like adding a SamCoreMLModel separately for the CoreML export using the current sam/repvit_sam/utils/onnx.py and keeping the SamOnnxModel the same as your PR.