diff --git a/src/spikeinterface/comparison/groundtruthstudy.py b/src/spikeinterface/comparison/groundtruthstudy.py
index df0b5296c0..4f9a0b2a14 100644
--- a/src/spikeinterface/comparison/groundtruthstudy.py
+++ b/src/spikeinterface/comparison/groundtruthstudy.py
@@ -165,7 +165,7 @@ def run_sorters(self, case_keys=None, engine="loop", engine_kwargs={}, keep=True
             sorting_exists = sorting_folder.exists()
 
             sorter_folder = self.folder / "sorters" / self.key_to_str(key)
-            sorter_folder_exists = sorting_folder.exists()
+            sorter_folder_exists = sorter_folder.exists()
 
             if keep:
                 if sorting_exists:
@@ -185,6 +185,9 @@ def run_sorters(self, case_keys=None, engine="loop", engine_kwargs={}, keep=True
                 if log_file.exists():
                     log_file.unlink()
 
+            if sorter_folder_exists:
+                shutil.rmtree(sorter_folder)
+
             params = self.cases[key]["run_sorter_params"].copy()
             # this ensure that sorter_name is given
             recording, _ = self.datasets[self.cases[key]["dataset"]]
diff --git a/src/spikeinterface/sorters/internal/spyking_circus2.py b/src/spikeinterface/sorters/internal/spyking_circus2.py
index 0c3b9f95d1..2c297662f4 100644
--- a/src/spikeinterface/sorters/internal/spyking_circus2.py
+++ b/src/spikeinterface/sorters/internal/spyking_circus2.py
@@ -114,7 +114,9 @@ def _run_from_folder(cls, sorter_output_folder, params, verbose):
 
         ## We get the labels for our peaks
         mask = peak_labels > -1
-        sorting = NumpySorting.from_times_labels(selected_peaks["sample_index"][mask], peak_labels[mask], sampling_rate)
+        sorting = NumpySorting.from_times_labels(
+            selected_peaks["sample_index"][mask], peak_labels[mask].astype(int), sampling_rate
+        )
         clustering_folder = sorter_output_folder / "clustering"
         if clustering_folder.exists():
             shutil.rmtree(clustering_folder)
diff --git a/src/spikeinterface/sortingcomponents/benchmark/benchmark_clustering.py b/src/spikeinterface/sortingcomponents/benchmark/benchmark_clustering.py
index d68b8e5449..bd413417bf 100644
--- a/src/spikeinterface/sortingcomponents/benchmark/benchmark_clustering.py
+++ b/src/spikeinterface/sortingcomponents/benchmark/benchmark_clustering.py
@@ -524,7 +524,7 @@ def plot_statistics(self, metric="cosine", annotations=True, detect_threshold=5)
             template_real = template_real.reshape(template_real.size, 1).T
 
             if metric == "cosine":
-                dist = sklearn.metrics.pairwise.cosine_similarity(template, template_real, metric).flatten().tolist()
+                dist = sklearn.metrics.pairwise.cosine_similarity(template, template_real).flatten().tolist()
             else:
                 dist = sklearn.metrics.pairwise_distances(template, template_real, metric).flatten().tolist()
             res += dist
diff --git a/src/spikeinterface/widgets/widget_list.py b/src/spikeinterface/widgets/widget_list.py
index ed77de6128..51e7208080 100644
--- a/src/spikeinterface/widgets/widget_list.py
+++ b/src/spikeinterface/widgets/widget_list.py
@@ -114,7 +114,7 @@
 plot_study_run_times = StudyRunTimesWidget
 plot_study_unit_counts = StudyUnitCountsWidget
 plot_study_performances = StudyPerformances
-plot_stufy_performances_vs_metrics = StudyPerformancesVsMetrics
+plot_study_performances_vs_metrics = StudyPerformancesVsMetrics
 
 
 def plot_timeseries(*args, **kwargs):