azure-ai-evaluation 1.8.0__py3-none-any.whl → 1.10.0__py3-none-any.whl

azure/ai/evaluation/_evaluators/_content_safety/_sexual.py

@@ -41,9 +41,9 @@ class SexualEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
 
     :param credential: The credential for connecting to Azure AI project. Required
     :type credential: ~azure.core.credentials.TokenCredential
-    :param azure_ai_project: The scope of the Azure AI project.
-        It contains subscription id, resource group, and project name.
-    :type azure_ai_project: ~azure.ai.evaluation.AzureAIProject
+    :param azure_ai_project: The Azure AI project, which can either be a string representing the project endpoint
+        or an instance of AzureAIProject. It contains subscription id, resource group, and project name.
+    :type azure_ai_project: Union[str, ~azure.ai.evaluation.AzureAIProject]
     :param threshold: The threshold for the Sexual evaluator. Default is 3.
     :type threshold: int
 
@@ -55,17 +55,17 @@ class SexualEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
             :language: python
             :dedent: 8
             :caption: Initialize and call a SexualEvaluator.
-    
+
     .. admonition:: Example using Azure AI Project URL:
-                
+
         .. literalinclude:: ../samples/evaluation_samples_evaluate_fdp.py
             :start-after: [START sexual_evaluator]
             :end-before: [END sexual_evaluator]
             :language: python
             :dedent: 8
-            :caption: Initialize and call SexualEvaluator using Azure AI Project URL in following format 
+            :caption: Initialize and call SexualEvaluator using Azure AI Project URL in following format
                 https://{resource_name}.services.ai.azure.com/api/projects/{project_name}
-    
+
     .. admonition:: Example with Threshold:
 
         .. literalinclude:: ../samples/evaluation_samples_threshold.py
@@ -76,8 +76,9 @@ class SexualEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
             :caption: Initialize with threshold and call a SexualEvaluator.
     """
 
-    id = "azureml://registries/azureml/models/Sexual-Content-Evaluator/versions/3"
+    id = "azureai://built-in/evaluators/sexual"
     """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
+    _OPTIONAL_PARAMS = ["query"]
 
     @override
     def __init__(
@@ -86,6 +87,7 @@ class SexualEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
         azure_ai_project,
         *,
         threshold: int = 3,
+        **kwargs,
     ):
         super().__init__(
             eval_metric=EvaluationMetrics.SEXUAL,
@@ -94,6 +96,7 @@ class SexualEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
             conversation_aggregation_type=_AggregationType.MAX,
             threshold=threshold,
             _higher_is_better=False,
+            **kwargs,
         )
 
     @overload
@@ -146,7 +149,7 @@ class SexualEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
             key "messages". Conversation turns are expected
             to be dictionaries with keys "content" and "role".
         :paramtype conversation: Optional[~azure.ai.evaluation.Conversation]
-        :return: The fluency score.
+        :return: The sexual score.
         :rtype: Union[Dict[str, Union[str, float]], Dict[str, Union[str, float, Dict[str, List[Union[str, float]]]]]]
         """
         return super().__call__(*args, **kwargs)

azure/ai/evaluation/_evaluators/_content_safety/_violence.py

@@ -41,9 +41,9 @@ class ViolenceEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
 
     :param credential: The credential for connecting to Azure AI project. Required
     :type credential: ~azure.core.credentials.TokenCredential
-    :param azure_ai_project: The scope of the Azure AI project.
-        It contains subscription id, resource group, and project name.
-    :type azure_ai_project: ~azure.ai.evaluation.AzureAIProject
+    :param azure_ai_project: The Azure AI project, which can either be a string representing the project endpoint
+        or an instance of AzureAIProject. It contains subscription id, resource group, and project name.
+    :type azure_ai_project: Union[str, ~azure.ai.evaluation.AzureAIProject]
     :param threshold: The threshold for the Violence evaluator. Default is 3.
     :type threshold: int
 
@@ -57,15 +57,15 @@ class ViolenceEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
             :caption: Initialize and call a ViolenceEvaluator.
 
     .. admonition:: Example using Azure AI Project URL:
-                
+
         .. literalinclude:: ../samples/evaluation_samples_evaluate_fdp.py
             :start-after: [START violence_evaluator]
             :end-before: [END violence_evaluator]
             :language: python
             :dedent: 8
-            :caption: Initialize and call ViolenceEvaluator using Azure AI Project URL in following format 
+            :caption: Initialize and call ViolenceEvaluator using Azure AI Project URL in following format
                 https://{resource_name}.services.ai.azure.com/api/projects/{project_name}
-    
+
     .. admonition:: Example:
 
         .. literalinclude:: ../samples/evaluation_samples_threshold.py
@@ -76,8 +76,9 @@ class ViolenceEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
             :caption: Initialize with threshold and call a ViolenceEvaluator.
     """
 
-    id = "azureml://registries/azureml/models/Violent-Content-Evaluator/versions/3"
+    id = "azureai://built-in/evaluators/violence"
     """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
+    _OPTIONAL_PARAMS = ["query"]
 
     @override
     def __init__(
@@ -86,6 +87,7 @@ class ViolenceEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
         azure_ai_project,
         *,
         threshold: int = 3,
+        **kwargs,
     ):
         super().__init__(
             eval_metric=EvaluationMetrics.VIOLENCE,
@@ -94,6 +96,7 @@ class ViolenceEvaluator(RaiServiceEvaluatorBase[Union[str, float]]):
             conversation_aggregation_type=_AggregationType.MAX,
             threshold=threshold,
             _higher_is_better=False,
+            **kwargs,
         )
 
     @overload

azure/ai/evaluation/_evaluators/_document_retrieval/__init__.py

@@ -4,8 +4,4 @@
 
 from ._document_retrieval import DocumentRetrievalEvaluator, RetrievalGroundTruthDocument, RetrievedDocument
 
-__all__ = [
-    "DocumentRetrievalEvaluator",
-    "RetrievalGroundTruthDocument",
-    "RetrievedDocument"
-]
+__all__ = ["DocumentRetrievalEvaluator", "RetrievalGroundTruthDocument", "RetrievedDocument"]

azure/ai/evaluation/_evaluators/_document_retrieval/_document_retrieval.py

@@ -14,9 +14,7 @@ RetrievalGroundTruthDocument = TypedDict(
     "RetrievalGroundTruthDocument", {"document_id": str, "query_relevance_label": int}
 )
 
-RetrievedDocument = TypedDict(
-    "RetrievedDocument", {"document_id": str, "relevance_score": float}
-)
+RetrievedDocument = TypedDict("RetrievedDocument", {"document_id": str, "relevance_score": float})
 
 
 class DocumentRetrievalEvaluator(EvaluatorBase):
@@ -33,15 +31,15 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
             :caption: Initialize and call a DocumentRetrievalEvaluator
 
     .. admonition:: Example using Azure AI Project URL:
-                
+
         .. literalinclude:: ../samples/evaluation_samples_evaluate_fdp.py
             :start-after: [START document_retrieval_evaluator]
             :end-before: [END document_retrieval_evaluator]
             :language: python
             :dedent: 8
-            :caption: Initialize and call DocumentRetrievalEvaluator using Azure AI Project URL in following format 
+            :caption: Initialize and call DocumentRetrievalEvaluator using Azure AI Project URL in following format
                 https://{resource_name}.services.ai.azure.com/api/projects/{project_name}
-    
+
     .. admonition:: Example with Threshold:
         .. literalinclude:: ../samples/evaluation_samples_threshold.py
             :start-after: [START threshold_document_retrieval_evaluator]
@@ -51,6 +49,9 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
             :caption: Initialize with threshold and call a DocumentRetrievalEvaluator.
     """
 
+    id = "azureai://built-in/evaluators/document_retrieval"
+    """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
+
     def __init__(
         self,
         *,
@@ -62,7 +63,7 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
         top1_relevance_threshold: Optional[float] = 50.0,
         top3_max_relevance_threshold: Optional[float] = 50.0,
         total_retrieved_documents_threshold: Optional[int] = 50,
-        total_ground_truth_documents_threshold: Optional[int] = 50
+        total_ground_truth_documents_threshold: Optional[int] = 50,
     ):
         super().__init__()
         self.k = 3
@@ -74,14 +75,10 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
             )
 
         if not isinstance(ground_truth_label_min, int):
-            raise EvaluationException(
-                "The ground truth label minimum must be an integer value."
-            )
+            raise EvaluationException("The ground truth label minimum must be an integer value.")
 
         if not isinstance(ground_truth_label_max, int):
-            raise EvaluationException(
-                "The ground truth label maximum must be an integer value."
-            )
+            raise EvaluationException("The ground truth label maximum must be an integer value.")
 
         self.ground_truth_label_min = ground_truth_label_min
         self.ground_truth_label_max = ground_truth_label_max
@@ -122,7 +119,7 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
     ) -> float:
         """NDCG (Normalized Discounted Cumulative Gain) calculated for the top K documents retrieved from a search query.
         NDCG measures how well a document ranking compares to an ideal document ranking given a list of ground-truth documents.
-        
+
         :param result_docs_groundtruth_labels: A list of retrieved documents' ground truth labels.
         :type result_docs_groundtruth_labels: List[int]
         :param ideal_docs_groundtruth_labels: A list of ideal documents' ground truth labels.
@@ -145,7 +142,7 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
     def _compute_xdcg(self, result_docs_groundtruth_labels: List[int]) -> float:
         """XDCG calculated for the top K documents retrieved from a search query.
         XDCG measures how objectively good are the top K documents, discounted by their position in the list.
-        
+
         :param result_docs_groundtruth_labels: A list of retrieved documents' ground truth labels.
         :type result_docs_groundtruth_labels: List[int]
         :return: The XDCG@K calculation result.
@@ -159,11 +156,7 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
             return math.pow(self.xdcg_discount_factor, rank - 1)
 
         ranks = list(range(1, self.k + 1))
-        xdcg_n = sum(
-            starmap(
-                calculate_xdcg_numerator, zip(result_docs_groundtruth_labels, ranks)
-            )
-        )
+        xdcg_n = sum(starmap(calculate_xdcg_numerator, zip(result_docs_groundtruth_labels, ranks)))
         xdcg_d = sum(map(calculate_xdcg_denominator, ranks))
 
         return xdcg_n / float(xdcg_d)
@@ -175,7 +168,7 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
     ) -> float:
         """Fidelity calculated over all documents retrieved from a search query.
         Fidelity measures how objectively good are all of the documents retrieved compared with all known good documents in the underlying data store.
-        
+
         :param result_docs_groundtruth_labels: A list of retrieved documents' ground truth labels.
         :type result_docs_groundtruth_labels: List[int]
         :param ideal_docs_groundtruth_labels: A list of ideal documents' ground truth labels.
@@ -196,25 +189,16 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
                 if label >= s:
                     label_counts[str(label)] += 1
 
-            sorted_label_counts = [
-                x[1] for x in sorted(label_counts.items(), key=lambda x: x[0])
-            ]
+            sorted_label_counts = [x[1] for x in sorted(label_counts.items(), key=lambda x: x[0])]
 
             # calculate weights
-            weights = [
-                (math.pow(2, i + 1) - 1)
-                for i in range(s, self.ground_truth_label_max + 1)
-            ]
+            weights = [(math.pow(2, i + 1) - 1) for i in range(s, self.ground_truth_label_max + 1)]
 
             # return weighted sum
             return sum(starmap(operator.mul, zip(sorted_label_counts, weights)))
 
-        weighted_sum_by_rating_results = calculate_weighted_sum_by_rating(
-            result_docs_groundtruth_labels
-        )
-        weighted_sum_by_rating_index = calculate_weighted_sum_by_rating(
-            ideal_docs_groundtruth_labels
-        )
+        weighted_sum_by_rating_results = calculate_weighted_sum_by_rating(result_docs_groundtruth_labels)
+        weighted_sum_by_rating_index = calculate_weighted_sum_by_rating(ideal_docs_groundtruth_labels)
 
         if weighted_sum_by_rating_index == 0:
             return math.nan
@@ -226,12 +210,16 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
 
         for metric_name, metric_value in metrics.items():
             if metric_name in self._threshold_metrics.keys():
-                result[f"{metric_name}_result"] = "pass" if metric_value >= self._threshold_metrics[metric_name] else "fail"
+                result[f"{metric_name}_result"] = (
+                    "pass" if metric_value >= self._threshold_metrics[metric_name] else "fail"
+                )
                 result[f"{metric_name}_threshold"] = self._threshold_metrics[metric_name]
                 result[f"{metric_name}_higher_is_better"] = True
 
             elif metric_name in self._threshold_holes.keys():
-                result[f"{metric_name}_result"] = "pass" if metric_value <= self._threshold_holes[metric_name] else "fail"
+                result[f"{metric_name}_result"] = (
+                    "pass" if metric_value <= self._threshold_holes[metric_name] else "fail"
+                )
                 result[f"{metric_name}_threshold"] = self._threshold_holes[metric_name]
                 result[f"{metric_name}_higher_is_better"] = False
 
@@ -256,8 +244,10 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
         # if the qrels are empty, no meaningful evaluation is possible
         if not retrieval_ground_truth:
             raise EvaluationException(
-                ("'retrieval_ground_truth' parameter must contain at least one item. "
-                 "Check your data input to be sure that each input record has ground truth defined.")
+                (
+                    "'retrieval_ground_truth' parameter must contain at least one item. "
+                    "Check your data input to be sure that each input record has ground truth defined."
+                )
             )
 
         qrels = []
@@ -277,9 +267,7 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
                 )
 
             if not isinstance(query_relevance_label, int):
-                raise EvaluationException(
-                    "Query relevance labels must be integer values."
-                )
+                raise EvaluationException("Query relevance labels must be integer values.")
 
             if query_relevance_label < self.ground_truth_label_min:
                 raise EvaluationException(
@@ -318,12 +306,8 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
                         )
                     )
 
-                if not isinstance(relevance_score, float) and not isinstance(
-                    relevance_score, int
-                ):
-                    raise EvaluationException(
-                        "Retrieved document relevance score must be a numerical value."
-                    )
+                if not isinstance(relevance_score, float) and not isinstance(relevance_score, int):
+                    raise EvaluationException("Retrieved document relevance score must be a numerical value.")
 
                 results.append(result)
 
@@ -368,24 +352,17 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
         results_lookup = {x["document_id"]: x["relevance_score"] for x in results}
 
         # sort each input set by label to get the ranking
-        qrels_sorted_by_rank = sorted(
-            qrels_lookup.items(), key=lambda x: x[1], reverse=True
-        )
-        results_sorted_by_rank = sorted(
-            results_lookup.items(), key=lambda x: x[1], reverse=True
-        )
+        qrels_sorted_by_rank = sorted(qrels_lookup.items(), key=lambda x: x[1], reverse=True)
+        results_sorted_by_rank = sorted(results_lookup.items(), key=lambda x: x[1], reverse=True)
 
         # find ground truth labels for the results set and ideal set
         result_docs_groundtruth_labels = [
-            qrels_lookup[doc_id] if doc_id in qrels_lookup else 0
-            for (doc_id, _) in results_sorted_by_rank
+            qrels_lookup[doc_id] if doc_id in qrels_lookup else 0 for (doc_id, _) in results_sorted_by_rank
         ]
         ideal_docs_groundtruth_labels = [label for (_, label) in qrels_sorted_by_rank]
 
         # calculate the proportion of result docs with no ground truth label (holes)
-        holes = self._compute_holes(
-            [x[0] for x in results_sorted_by_rank], [x[0] for x in qrels_sorted_by_rank]
-        )
+        holes = self._compute_holes([x[0] for x in results_sorted_by_rank], [x[0] for x in qrels_sorted_by_rank])
         holes_ratio = holes / float(len(results))
 
         # if none of the retrieved docs are labeled, report holes only
@@ -412,12 +389,8 @@ class DocumentRetrievalEvaluator(EvaluatorBase):
                 result_docs_groundtruth_labels[: self.k],
                 ideal_docs_groundtruth_labels[: self.k],
             ),
-            f"xdcg@{self.k}": self._compute_xdcg(
-                result_docs_groundtruth_labels[: self.k]
-            ),
-            "fidelity": self._compute_fidelity(
-                result_docs_groundtruth_labels, ideal_docs_groundtruth_labels
-            ),
+            f"xdcg@{self.k}": self._compute_xdcg(result_docs_groundtruth_labels[: self.k]),
+            "fidelity": self._compute_fidelity(result_docs_groundtruth_labels, ideal_docs_groundtruth_labels),
             "top1_relevance": result_docs_groundtruth_labels[0],
             "top3_max_relevance": max(result_docs_groundtruth_labels[: self.k]),
             "holes": holes,

azure/ai/evaluation/_evaluators/_eci/_eci.py

@@ -22,9 +22,9 @@ class ECIEvaluator(RaiServiceEvaluatorBase):
 
     :param credential: The credential for connecting to Azure AI project. Required
     :type credential: ~azure.core.credentials.TokenCredential
-    :param azure_ai_project: The scope of the Azure AI project.
-        It contains subscription id, resource group, and project name.
-    :type azure_ai_project: ~azure.ai.evaluation.AzureAIProject
+    :param azure_ai_project: The Azure AI project, which can either be a string representing the project endpoint
+        or an instance of AzureAIProject. It contains subscription id, resource group, and project name.
+    :type azure_ai_project: Union[str, ~azure.ai.evaluation.AzureAIProject]
     :return: Whether or not ECI was found in the response without a disclaimer, with AI-generated reasoning
     :rtype: Dict[str, str]
 
@@ -52,17 +52,20 @@ class ECIEvaluator(RaiServiceEvaluatorBase):
 
     id = "eci"
     """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
+    _OPTIONAL_PARAMS = ["query"]
 
     @override
     def __init__(
         self,
         credential,
         azure_ai_project,
+        **kwargs,
     ):
         super().__init__(
             eval_metric=_InternalEvaluationMetrics.ECI,
             azure_ai_project=azure_ai_project,
             credential=credential,
+            **kwargs,
         )
 
     @overload

azure/ai/evaluation/_evaluators/_f1_score/_f1_score.py

@@ -39,15 +39,15 @@ class F1ScoreEvaluator(EvaluatorBase):
             :caption: Initialize and call an F1ScoreEvaluator.
 
     .. admonition:: Example using Azure AI Project URL:
-                
+
         .. literalinclude:: ../samples/evaluation_samples_evaluate_fdp.py
             :start-after: [START f1_score_evaluator]
             :end-before: [END f1_score_evaluator]
             :language: python
             :dedent: 8
-            :caption: Initialize and call F1ScoreEvaluator using Azure AI Project URL in following format 
+            :caption: Initialize and call F1ScoreEvaluator using Azure AI Project URL in following format
                 https://{resource_name}.services.ai.azure.com/api/projects/{project_name}
-    
+
     .. admonition:: Example with Threshold:
 
         .. literalinclude:: ../samples/evaluation_samples_threshold.py
@@ -58,7 +58,7 @@ class F1ScoreEvaluator(EvaluatorBase):
             :caption: Initialize with threshold and call an F1ScoreEvaluator.
     """
 
-    id = "azureml://registries/azureml/models/F1Score-Evaluator/versions/3"
+    id = "azureai://built-in/evaluators/f1_score"
     """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
 
     def __init__(self, *, threshold=0.5):
@@ -147,7 +147,7 @@ class F1ScoreEvaluator(EvaluatorBase):
             if f1_result <= self._threshold:
                 binary_result = True
         return {
-            "f1_score": f1_result, 
+            "f1_score": f1_result,
             "f1_result": EVALUATION_PASS_FAIL_MAPPING[binary_result],
             "f1_threshold": self._threshold,
         }

azure/ai/evaluation/_evaluators/_fluency/_fluency.py

@@ -45,7 +45,7 @@ class FluencyEvaluator(PromptyEvaluatorBase[Union[str, float]]):
             :caption: Initialize with threshold and call a FluencyEvaluator.
 
     .. admonition:: Example using Azure AI Project URL:
-                
+
         .. literalinclude:: ../samples/evaluation_samples_evaluate_fdp.py
             :start-after: [START fluency_evaluator]
             :end-before: [END fluency_evaluator]
@@ -64,7 +64,7 @@ class FluencyEvaluator(PromptyEvaluatorBase[Union[str, float]]):
     _PROMPTY_FILE = "fluency.prompty"
     _RESULT_KEY = "fluency"
 
-    id = "azureml://registries/azureml/models/Fluency-Evaluator/versions/4"
+    id = "azureai://built-in/evaluators/fluency"
     """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
 
     @override
@@ -78,7 +78,7 @@ class FluencyEvaluator(PromptyEvaluatorBase[Union[str, float]]):
             prompty_file=prompty_path,
             result_key=self._RESULT_KEY,
             threshold=threshold,
-            _higher_is_better=self._higher_is_better
+            _higher_is_better=self._higher_is_better,
         )
 
     @overload

azure/ai/evaluation/_evaluators/_gleu/_gleu.py

@@ -34,7 +34,7 @@ class GleuScoreEvaluator(EvaluatorBase):
             :language: python
             :dedent: 8
             :caption: Initialize and call a GleuScoreEvaluator.
-    
+
     .. admonition:: Example with Threshold:
 
         .. literalinclude:: ../samples/evaluation_samples_threshold.py
@@ -45,17 +45,17 @@ class GleuScoreEvaluator(EvaluatorBase):
             :caption: Initialize with threshold and call a GleuScoreEvaluator.
 
     .. admonition:: Example using Azure AI Project URL:
-                
+
         .. literalinclude:: ../samples/evaluation_samples_evaluate_fdp.py
             :start-after: [START gleu_score_evaluator]
             :end-before: [END gleu_score_evaluator]
             :language: python
             :dedent: 8
-            :caption: Initialize and call GleuScoreEvaluator using Azure AI Project URL in the following format 
+            :caption: Initialize and call GleuScoreEvaluator using Azure AI Project URL in the following format
                 https://{resource_name}.services.ai.azure.com/api/projects/{project_name}
     """
 
-    id = "azureml://registries/azureml/models/Gleu-Score-Evaluator/versions/3"
+    id = "azureai://built-in/evaluators/gleu_score"
     """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
 
     @override

azure/ai/evaluation/_evaluators/_groundedness/_groundedness.py

@@ -12,9 +12,13 @@ from azure.ai.evaluation._model_configurations import Conversation
 from ..._common.utils import construct_prompty_model_config, validate_model_config
 
 try:
-    from ..._user_agent import USER_AGENT
+    from ..._user_agent import UserAgentSingleton
 except ImportError:
-    USER_AGENT = "None"
+
+    class UserAgentSingleton:
+        @property
+        def value(self) -> str:
+            return "None"
 
 
 class GroundednessEvaluator(PromptyEvaluatorBase[Union[str, float]]):
@@ -35,7 +39,7 @@ class GroundednessEvaluator(PromptyEvaluatorBase[Union[str, float]]):
         ~azure.ai.evaluation.OpenAIModelConfiguration]
     :param threshold: The threshold for the groundedness evaluator. Default is 3.
     :type threshold: int
-        
+
     .. admonition:: Example:
 
         .. literalinclude:: ../samples/evaluation_samples_evaluate.py
@@ -54,13 +58,13 @@ class GroundednessEvaluator(PromptyEvaluatorBase[Union[str, float]]):
             :caption: Initialize with threshold and call a GroundednessEvaluator.
 
     .. admonition:: Example using Azure AI Project URL:
-        
+
         .. literalinclude:: ../samples/evaluation_samples_evaluate_fdp.py
             :start-after: [START groundedness_evaluator]
             :end-before: [END groundedness_evaluator]
             :language: python
             :dedent: 8
-            :caption: Initialize and call GroundednessEvaluator using Azure AI Project URL in the following format 
+            :caption: Initialize and call GroundednessEvaluator using Azure AI Project URL in the following format
                 https://{resource_name}.services.ai.azure.com/api/projects/{project_name}
 
     .. note::
@@ -75,7 +79,7 @@ class GroundednessEvaluator(PromptyEvaluatorBase[Union[str, float]]):
     _RESULT_KEY = "groundedness"
     _OPTIONAL_PARAMS = ["query"]
 
-    id = "azureml://registries/azureml/models/Groundedness-Evaluator/versions/4"
+    id = "azureai://built-in/evaluators/groundedness"
     """Evaluator identifier, experimental and to be used only with evaluation in cloud."""
 
     @override
@@ -89,7 +93,7 @@ class GroundednessEvaluator(PromptyEvaluatorBase[Union[str, float]]):
             prompty_file=prompty_path,
             result_key=self._RESULT_KEY,
             threshold=threshold,
-            _higher_is_better=self._higher_is_better
+            _higher_is_better=self._higher_is_better,
         )
         self._model_config = model_config
         self.threshold = threshold
@@ -165,7 +169,7 @@ class GroundednessEvaluator(PromptyEvaluatorBase[Union[str, float]]):
             prompty_model_config = construct_prompty_model_config(
                 validate_model_config(self._model_config),
                 self._DEFAULT_OPEN_API_VERSION,
-                USER_AGENT,
+                UserAgentSingleton().value,
             )
             self._flow = AsyncPrompty.load(source=self._prompty_file, model=prompty_model_config)