diff --git a/langtest/tasks/task.py b/langtest/tasks/task.py
index c14a568f4..cb2543eaf 100644
--- a/langtest/tasks/task.py
+++ b/langtest/tasks/task.py
@@ -26,6 +26,7 @@
     SensitivitySample,
     LLMAnswerSample,
     CrowsPairsSample,
+    StereoSetSample,
 )
 from langtest.utils.custom_types.predictions import NERPrediction
 
@@ -707,3 +708,65 @@ def create_sample(
             mask1=row_data[column_mapper[mask1]],
             mask2=row_data[column_mapper[mask2]],
         )
+
+
+class StereosetTask(BaseTask):
+    """StereoSet task."""
+
+    _name = "stereoset"
+    _default_col = {
+        "text": ["text", "sentence"],
+        "mask1": ["mask1"],
+        "mask2": ["mask2"],
+    }
+    sample_class = StereoSetSample
+
+    def create_sample(
+        cls,
+        row_data: dict,
+        bias_type: str = "bias_type",
+        test_type: str = "type",
+        target_column: str = "target",
+        context: str = "context",
+        sent_stereo: str = "stereotype",
+        sent_antistereo: str = "anti-stereotype",
+        sent_unrelated: str = "unrelated",
+        *args,
+        **kwargs,
+    ) -> StereoSetSample:
+        """Create a sample."""
+        keys = list(row_data.keys())
+        if set(
+            [
+                bias_type,
+                test_type,
+                target_column,
+                context,
+                sent_stereo,
+                sent_antistereo,
+                sent_unrelated,
+            ]
+        ).intersection(set(keys)):
+            # if the column names are provided, use them directly
+            column_mapper = {
+                bias_type: bias_type,
+                test_type: test_type,
+                target_column: target_column,
+                context: context,
+                sent_stereo: sent_stereo,
+                sent_antistereo: sent_antistereo,
+                sent_unrelated: sent_unrelated,
+            }
+        else:
+            # auto-detect the default column names from the row_data
+            column_mapper = cls.column_mapping(keys)
+
+        return StereoSetSample(
+            test_type=row_data[column_mapper[test_type]],
+            bias_type=row_data[column_mapper[bias_type]],
+            target=row_data[column_mapper[target_column]],
+            context=row_data[column_mapper[context]],
+            sent_stereo=row_data[column_mapper[sent_stereo]],
+            sent_antistereo=row_data[column_mapper[sent_antistereo]],
+            sent_unrelated=row_data[column_mapper[sent_unrelated]],
+        )