Save .slp file after inference

sleap_nn/inference/predictors.py

@@ -363,7 +363,7 @@ def predict(
             # Create SLEAP data structures from the predictions.
             pred_labels = self._make_labeled_frames_from_generator(generator)
             if save_path:
-                sio.io.slp.write_labels(save_path, pred_labels)
+                pred_labels.save(Path(save_path).as_posix())
             return pred_labels
 
         else:

sleap_nn/training/model_trainer.py

@@ -190,6 +190,8 @@ def __init__(
             else:
                 self.crop_hw = self.crop_hw[0]
 
+        OmegaConf.save(config=self.config, f=f"{self.dir_path}/training_config.yaml")
+
     def _create_data_loaders_torch_dataset(self):
         """Create a torch DataLoader for train, validation and test sets using the data_config."""
         train_labels = sio.load_slp(self.config.data_config.train_labels_path)

tests/inference/test_predictors.py

@@ -13,6 +13,7 @@ def test_topdown_predictor(
     minimal_instance_ckpt,
     minimal_instance_centroid_ckpt,
     minimal_instance_bottomup_ckpt,
+    tmp_path,
 ):
     """Test TopDownPredictor class for running inference on centroid and centered instance models."""
     # for centered instance model
@@ -25,12 +26,15 @@ def test_topdown_predictor(
         return_confmaps=False,
         make_labels=True,
         peak_threshold=0.0,
+        save_path=f"{tmp_path}/test.pkg.slp",
     )
     assert isinstance(pred_labels, sio.Labels)
     assert len(pred_labels) == 1
     assert len(pred_labels[0].instances) == 2
     lf = pred_labels[0]
 
+    assert Path(f"{tmp_path}/test.pkg.slp").exists
+
     # check if the predicted labels have same video and skeleton as the ground truth labels
     gt_labels = sio.load_slp(minimal_instance)
     gt_lf = gt_labels[0]