Merge pull request #207 from EthoML/dev

dev

Author: luiztauffer

.clinerules

@@ -1,3 +1,11 @@
+# v0.12.1
+
+### Fixes
+
+- Add Apple Silicon (MPS) GPU support
+- Centralize device detection in `get_device()` utility
+
+
 # v0.12.0
 
 ### Fixes

CHANGELOG.md

@@ -7,7 +7,7 @@ title: model.evaluate
 
 #### logger
 
-#### use\_gpu
+#### DEVICE
 
 #### create\_reconstruction\_plot
 
@@ -48,7 +48,6 @@ Saves the plot to:
 
 ```python
 def eval_temporal(config: dict,
-                  use_gpu: bool,
                   model_name: str,
                   fixed: bool,
                   snapshot: Optional[str] = None,
@@ -60,7 +59,6 @@ Evaluate the temporal aspects of the trained model.
 **Parameters**
 
 * **config** (`dict`): Configuration dictionary.
-* **use_gpu** (`bool`): Flag indicating whether to use GPU for evaluation.
 * **model_name** (`str`): Name of the model.
 * **fixed** (`bool`): Flag indicating whether the data is fixed or not.
 * **snapshot** (`str, optional`): Path to the model snapshot. Defaults to None.

docs/vame-docs-app/docs/reference/model/evaluate.md

@@ -19,7 +19,7 @@ Log every N batches
 
 Set to True to log parameter histograms
 
-#### use\_gpu
+#### DEVICE
 
 #### reconstruction\_loss
 

docs/vame-docs-app/docs/reference/model/rnn_vae.md

@@ -19,12 +19,20 @@ Gets the VAME package version from pyproject.toml.
 
 * `str`: The version string.
 
-#### check\_torch\_device
+#### get\_device
 
 ```python
-def check_torch_device() -> bool
+def get_device() -> torch.device
 ```
 
+Detect the best available compute device.
+
+Checks for CUDA, then MPS (Apple Silicon), then falls back to CPU.
+
+**Returns**
+
+* `torch.device`: The selected device.
+
 #### \_convert\_enums\_to\_values
 
 ```python