Bugfix/mps acceleration support

ahcore/transforms/pre_transforms.py

@@ -232,6 +232,10 @@ def __call__(self, sample: DlupDatasetSample) -> dict[str, DlupDatasetSample]:
             roi = sample["annotation_data"]["roi"]
             sample["roi"] = torch.from_numpy(roi[np.newaxis, ...]).float()
 
+        sample["mpp"] = torch.tensor(
+            sample["mpp"], dtype=torch.float32 if torch.backends.mps.is_available() else torch.float64
+        )
+
         return sample
 
     def __repr__(self) -> str:

config/trainer/mps.yaml

@@ -0,0 +1,16 @@
+_target_: pytorch_lightning.Trainer
+
+# When using MPS acceleration set environment variable `PYTORCH_ENABLE_MPS_FALLBACK=1` in .env
+accelerator: mps
+devices: 1
+max_epochs: 1000
+num_nodes: 1
+
+# number of validation steps to execute at the beginning of the training
+num_sanity_val_steps: 0
+#log_every_n_steps: 2
+
+# if a smaller grid is used, the val check interval should be smaller
+#val_check_interval: 2  # Used if you want to check val more than once per epoch
+check_val_every_n_epoch: 1  # Used if you want to check val less than once per epoch
+accumulate_grad_batches: 1