set source on workflow execution inference requests

docs/inference_helpers/inference_sdk.md

@@ -553,6 +553,8 @@ The following fields are passed to API
   `disable_preproc_static_crop` to alter server-side pre-processing
 - `disable_active_learning` to prevent Active Learning feature from registering the datapoint (can be useful for
   instance while testing model)
+- `source` Optional string to set a "source" attribute on the inference call; if using model monitoring, this will get logged with the inference request so you can filter/query inference requests coming from a particular source. e.g. to identify which application, system, or deployment is making the request.
+- `source_info` Optional string to set additional "source_info" attribute on the inference call; e.g. to identify a sub component in an app.
 
 The following fields are passed to API
 
@@ -571,6 +573,8 @@ The following fields are passed to API
 - `disable_preproc_auto_orientation`, `disable_preproc_contrast`, `disable_preproc_grayscale`,
   `disable_preproc_static_crop` to alter server-side pre-processing
 - `disable_active_learning` to prevent Active Learning feature from registering the datapoint (can be useful for instance while testing model)
+- `source` Optional string to set a "source" attribute on the inference call; if using model monitoring, this will get logged with the inference request so you can filter/query inference requests coming from a particular source. e.g. to identify which application, system, or deployment is making the request.
+- `source_info` Optional string to set additional "source_info" attribute on the inference call; e.g. to identify a sub component in an app.
 
 ### Classification model in `v1` mode:
 
@@ -589,6 +593,9 @@ The following fields are passed to API
   `disable_preproc_static_crop` to alter server-side pre-processing
 * `disable_active_learning` to prevent Active Learning feature from registering the datapoint (can be useful for instance while testing model)
 
+- `source` Optional string to set a "source" attribute on the inference call; if using model monitoring, this will get logged with the inference request so you can filter/query inference requests coming from a particular source. e.g. to identify which application, system, or deployment is making the request.
+- `source_info` Optional string to set additional "source_info" attribute on the inference call; e.g. to identify a sub component in an app.
+
 ### Object detection model in `v1` mode:
 
 - `visualize_predictions`: flag to enable / disable visualisation
@@ -605,6 +612,8 @@ The following fields are passed to API
   `disable_preproc_static_crop` to alter server-side pre-processing
 - `disable_active_learning` to prevent Active Learning feature from registering the datapoint (can be useful for
   instance while testing model)
+- `source` Optional string to set a "source" attribute on the inference call; if using model monitoring, this will get logged with the inference request so you can filter/query inference requests coming from a particular source. e.g. to identify which application, system, or deployment is making the request.
+- `source_info` Optional string to set additional "source_info" attribute on the inference call; e.g. to identify a sub component in an app.
 
 ### Keypoints detection model in `v1` mode:
 
@@ -624,6 +633,8 @@ The following fields are passed to API
   `disable_preproc_static_crop` to alter server-side pre-processing
 - `disable_active_learning` to prevent Active Learning feature from registering the datapoint (can be useful for
   instance while testing model)
+- `source` Optional string to set a "source" attribute on the inference call; if using model monitoring, this will get logged with the inference request so you can filter/query inference requests coming from a particular source. e.g. to identify which application, system, or deployment is making the request.
+- `source_info` Optional string to set additional "source_info" attribute on the inference call; e.g. to identify a sub component in an app.
 
 ### Instance segmentation model in `v1` mode:
 
@@ -643,6 +654,8 @@ The following fields are passed to API
 - `tradeoff_factor`
 - `disable_active_learning` to prevent Active Learning feature from registering the datapoint (can be useful for
   instance while testing model)
+- `source` Optional string to set a "source" attribute on the inference call; if using model monitoring, this will get logged with the inference request so you can filter/query inference requests coming from a particular source. e.g. to identify which application, system, or deployment is making the request.
+- `source_info` Optional string to set additional "source_info" attribute on the inference call; e.g. to identify a sub component in an app.
 
 ### Configuration of client
 

inference/enterprise/workflows/complier/steps_executors/models.py

@@ -187,6 +187,7 @@ def construct_classification_request(
         image=image,
         confidence=resolve(step.confidence),
         disable_active_learning=resolve(step.disable_active_learning),
+        source="workflow-execution",
     )
 
 
@@ -213,6 +214,7 @@ def construct_object_detection_request(
         iou_threshold=resolve(step.iou_threshold),
         max_detections=resolve(step.max_detections),
         max_candidates=resolve(step.max_candidates),
+        source="workflow-execution",
     )
 
 
@@ -241,6 +243,7 @@ def construct_instance_segmentation_request(
         max_candidates=resolve(step.max_candidates),
         mask_decode_mode=resolve(step.mask_decode_mode),
         tradeoff_factor=resolve(step.tradeoff_factor),
+        source="workflow-execution",
     )
 
 
@@ -268,6 +271,7 @@ def construct_keypoints_detection_request(
         max_detections=resolve(step.max_detections),
         max_candidates=resolve(step.max_candidates),
         keypoint_confidence=resolve(step.keypoint_confidence),
+        source="workflow-execution",
     )
 
 

inference_sdk/http/entities.py

@@ -86,6 +86,8 @@ class InferenceConfiguration:
     disable_active_learning: bool = False
     max_concurrent_requests: int = 1
     max_batch_size: int = 1
+    source: Optional[str] = None
+    source_info: Optional[str] = None
 
     @classmethod
     def init_default(cls) -> "InferenceConfiguration":
@@ -130,6 +132,8 @@ def to_object_detection_parameters(self) -> Dict[str, Any]:
             ("stroke_width", "visualization_stroke_width"),
             ("visualize_predictions", "visualize_predictions"),
             ("disable_active_learning", "disable_active_learning"),
+            ("source", "source"),
+            ("source_info", "source_info"),
         ]
         return get_non_empty_attributes(
             source_object=self,
@@ -141,6 +145,8 @@ def to_instance_segmentation_parameters(self) -> Dict[str, Any]:
         parameters_specs = [
             ("mask_decode_mode", "mask_decode_mode"),
             ("tradeoff_factor", "tradeoff_factor"),
+            ("source", "source"),
+            ("source_info", "source_info"),
         ]
         for internal_name, external_name in parameters_specs:
             parameters[external_name] = getattr(self, internal_name)
@@ -156,6 +162,8 @@ def to_classification_parameters(self) -> Dict[str, Any]:
             ("visualize_predictions", "visualize_predictions"),
             ("stroke_width", "visualization_stroke_width"),
             ("disable_active_learning", "disable_active_learning"),
+            ("source", "source"),
+            ("source_info", "source_info"),
         ]
         return get_non_empty_attributes(
             source_object=self,
@@ -180,6 +188,8 @@ def to_legacy_call_parameters(self) -> Dict[str, Any]:
             ("disable_preproc_grayscale", "disable_preproc_grayscale"),
             ("disable_preproc_static_crop", "disable_preproc_static_crop"),
             ("disable_active_learning", "disable_active_learning"),
+            ("source", "source"),
+            ("source_info", "source_info"),
         ]
         return get_non_empty_attributes(
             source_object=self,

tests/inference_sdk/unit_tests/http/test_client.py

@@ -230,7 +230,7 @@ async def example() -> None:
 def test_setting_configuration_statically() -> None:
     # given
     http_client = InferenceHTTPClient(api_key="my-api-key", api_url="https://some.com")
-    configuration = InferenceConfiguration(visualize_labels=True)
+    configuration = InferenceConfiguration(visualize_labels=True, source="source-test")
 
     # when
     previous_configuration = http_client.inference_configuration

tests/inference_sdk/unit_tests/http/test_entities.py

@@ -28,6 +28,8 @@
     visualize_labels=True,
     iou_threshold=0.7,
     disable_active_learning=True,
+    source="config-test",
+    source_info="config-test-source-info",
 )
 
 
@@ -55,6 +57,29 @@ def test_get_non_empty_attributes() -> None:
     }
 
 
+def test_source_attributes() -> None:
+    # given
+    reference_object = InferenceConfiguration(
+        source="source-test",
+        source_info="source-info-test",
+    )
+
+    # when
+    result = get_non_empty_attributes(
+        source_object=reference_object,
+        specification=[
+            ("source", "A"),
+            ("source_info", "B"),
+        ],
+    )
+
+    # then
+    assert result == {
+        "A": "source-test",
+        "B": "source-info-test",
+    }
+
+
 def test_to_api_call_parameters_for_api_v0() -> None:
     # when
     result = REFERENCE_IMAGE_CONFIGURATION.to_api_call_parameters(
@@ -78,6 +103,8 @@ def test_to_api_call_parameters_for_api_v0() -> None:
         "disable_preproc_grayscale": True,
         "disable_preproc_static_crop": False,
         "disable_active_learning": True,
+        "source": "config-test",
+        "source_info": "config-test-source-info",
     }
 
 
@@ -98,6 +125,8 @@ def test_to_api_call_parameters_for_api_v1_classification() -> None:
         "disable_preproc_grayscale": True,
         "disable_preproc_static_crop": False,
         "disable_active_learning": True,
+        "source": "config-test",
+        "source_info": "config-test-source-info",
     }
 
 
@@ -124,4 +153,6 @@ def test_to_api_call_parameters_for_api_v1_object_detection() -> None:
         "visualize_predictions": False,
         "visualization_stroke_width": 1,
         "disable_active_learning": True,
+        "source": "config-test",
+        "source_info": "config-test-source-info",
     }