Merge branch 'main' into llama

lm_eval/tasks/moral_stories/utils.py

+import datasets
+
+
+def process_docs(dataset: datasets.Dataset) -> datasets.Dataset:
+    def _process_doc(doc):
+        ctx = (
+            doc["norm"].capitalize()
+            + " "
+            + doc["situation"].capitalize()
+            + " "
+            + doc["intention"].capitalize()
+        )
+        choices = [doc["moral_action"], doc["immoral_action"]]
+        out_doc = {
+            "query": ctx,
+            "choices": choices,
+            "label": 0,
+        }
+        return out_doc
+
+    return dataset.map(_process_doc)