Merge remote-tracking branch 'origin/main'

evals/metrics/ragas/ragas.py

@@ -33,7 +33,7 @@ def __init__(
         self.model = model
         self.embeddings = embeddings
         self.metrics = metrics
-        
+
         # self.validated_list = [
         #     "answer_correctness",
         #     "answer_relevancy",
@@ -53,11 +53,12 @@ def measure(self, test_case: Dict):
         try:
             from ragas import evaluate
             from ragas.metrics import ALL_METRICS
-            self.metric_names = [metric.__class__.__name__ for metric in ALL_METRICS] 
-            self.metric_names = [re.sub(r'(?<!^)(?=[A-Z])', '_', name).lower() for name in self.metric_names]
+
+            self.metric_names = [metric.__class__.__name__ for metric in ALL_METRICS]
+            self.metric_names = [re.sub(r"(?<!^)(?=[A-Z])", "_", name).lower() for name in self.metric_names]
             self.metric_names = list(set(self.metric_names))
             # Note - summarization score metric is not working with best open-source LLMs
-            # Note - which is why we are removing it from our offering at the moment. 
+            # Note - which is why we are removing it from our offering at the moment.
             self.metric_names.remove("summarization_score")
             self.metric_instances = {}
             for metric in self.metric_names:
@@ -141,14 +142,15 @@ def measure(self, test_case: Dict):
             "response": "answer",
             "reference": "ground_truth",
             "retrieved_contexts": "contexts",
-            "reference_contexts": "reference_contexts"
+            "reference_contexts": "reference_contexts",
         }
         for metric in self.metrics:
             if hasattr(metric, "_required_columns"):
                 for column in list(metric._required_columns.values())[0]:
                     _required_columns.add(column_map[column])
             elif hasattr(metric, "evaluation_mode"):
                 from ragas.metrics.base import get_required_columns
+
                 for column in get_required_columns(metric.evaluation_mode):
                     _required_columns.add(column)
             else:

tests/test_ragas.py

@@ -30,15 +30,13 @@ def test_ragas(self):
 
         embeddings = HuggingFaceBgeEmbeddings(model_name="BAAI/bge-base-en-v1.5")
 
-        metric = RagasMetric(threshold=0.5, 
-                                model=f"http://{host_ip}:{port}", 
-                                embeddings=embeddings)
+        metric = RagasMetric(threshold=0.5, model=f"http://{host_ip}:{port}", embeddings=embeddings)
         test_case = {
             "question": ["What if these shoes don't fit?"],
             "answer": [actual_output],
             "ground_truth": [expected_output],
             "contexts": [retrieval_context],
-            "reference_contexts": [reference_context]
+            "reference_contexts": [reference_context],
         }
 
         metric.measure(test_case)