chore: import upstream snapshot with attribution

examples/flows/evaluation/eval-multi-turn-metrics/validate_input.py

@@ -0,0 +1,30 @@
+from promptflow.core import tool
+
+
+# Validate the metric's inputs.
+def is_valid(metric):
+    return True
+
+
+@tool
+def validate_input(chat_history: list, selected_metrics: dict) -> dict:
+    dict_metric_required_fields = {"answer_relevance": set(["question", "answer"]),
+                                   "conversation_quality": set(["question", "answer"]),
+                                   "creativity": set(["question", "answer"]),
+                                   "grounding": set(["answer", "context"])}
+    actual_input_cols = set()
+    for item in chat_history:
+        actual_input_cols.update(set(item["inputs"].keys()))
+        actual_input_cols.update(set(item["outputs"].keys()))
+        break
+
+    data_validation = selected_metrics
+    for metric in selected_metrics:
+        if selected_metrics[metric]:
+            metric_required_fields = dict_metric_required_fields[metric]
+            if metric_required_fields <= actual_input_cols:
+                data_validation[metric] = True
+            else:
+                print("this path")
+                data_validation[metric] = False
+    return data_validation