Snap

vps_clustering_benchmark.py

@@ -64,7 +64,7 @@ class VPClusteringBenchmark(datasets.GeneratorBasedBuilder):
                 "speaker": datasets.Value("string"),
                 "duration": datasets.Value("float32"),
                 "segment_clean": datasets.Value("bool"),
-                "vp": datasets.features.Array(shape=(192,), dtype="float32")
+                "vp": datasets.Sequence(datasets.Value("float32"))
             })
         return datasets.DatasetInfo(
             description=description,
@@ -95,5 +95,5 @@ class VPClusteringBenchmark(datasets.GeneratorBasedBuilder):
                 "speaker": row["speaker"],
                 "duration": row["duration"],
                 "segment_clean": row["segment_clean"],
-                "vp": np.array(row["vp"], dtype=np.float32)  # Ensure vp is a NumPy array
+                "vp": row["vp"].tolist() if isinstance(row["vp"], np.ndarray) else row["vp"]
             }