MacM1 mps error

[darren-pei]

Hey

Has anyone been successful in getting detectron2 to run on Mac mps? I've verified that pytorch is working well, but I've getting the following error:

[08/06 14:26:36 d2.data.dataset_mapper]: [DatasetMapper] Augmentations used in training: [ResizeShortestEdge(short_edge_length=(640, 672, 704, 736, 768, 800), max_size=1333, sample_style='choice'), RandomFlip()]
[08/06 14:26:36 d2.data.build]: Using training sampler TrainingSampler
[08/06 14:26:36 d2.data.common]: Serializing 399768 elements to byte tensors and concatenating them all ...
[08/06 14:26:38 d2.data.common]: Serialized dataset takes 255.14 MiB
WARNING [08/06 14:26:39 d2.solver.build]: SOLVER.STEPS contains values larger than SOLVER.MAX_ITER. These values will be ignored.
[08/06 14:26:39 d2.engine.train_loop]: Starting training from iteration 0
1.12.0 False
loc("mps_subtract"("(mpsFileLoc): /AppleInternal/Library/BuildRoots/20d6c351-ee94-11ec-bcaf-7247572f23b4/Library/Caches/com.apple.xbs/Sources/MetalPerformanceShadersGraph/mpsgraph/MetalPerformanceShadersGraph/Core/Files/MPSGraphUtilities.mm":219:0)): error: input types 'tensor<3x994x704xi8>' and 'tensor<3x1x1xf32>' are not broadcast compatible
LLVM ERROR: Failed to infer result type(s).