set up model test metrics that can either be pulled from Training or …

…from Inference Runs

applications/aws_dashboard/pages/models/layout.py

@@ -49,7 +49,7 @@ def models_layout(
                     dbc.Col(
                         [
                             dbc.Row(
-                                html.H3("Model Metrics", id="model_metrics_header"),
+                                html.H3("Model Testing", id="model_metrics_header"),
                                 style={"padding": "30px 0px 10px 0px"},
                             ),
                             dbc.Row(

src/sageworks/artifacts/models/model.py

@@ -128,6 +128,16 @@ def regression_predictions(self) -> Union[pd.DataFrame, None]:
             df = self._pull_s3_model_artifacts(s3_path)
             return df
 
+    def _pull_inference_metadata(self) -> Union[dict, None]:
+        """Internal: Retrieve the inference metadata for this model
+        Returns:
+            dict: Dictionary of the inference metadata (might be None)
+        Notes:
+            Basically when the inference was run, name of the dataset, the MD5, etc
+        """
+        s3_path = f"{self.model_inference_path}/{self.model_name}/inference_meta.csv"
+        return wr.s3.read_json(s3_path)
+
     def _pull_inference_metrics(self) -> Union[pd.DataFrame, None]:
         """Internal: Retrieve the inference model metrics for this model
         Returns:
@@ -230,6 +240,9 @@ def details(self, recompute=False) -> dict:
         details["confusion_matrix"] = self.confusion_matrix()
         details["regression_predictions"] = self.regression_predictions()
 
+        # Grab the inference metadata
+        details["inference_meta"] = self._pull_inference_metadata()
+
         # Cache the details
         self.data_storage.set(storage_key, details)
 

src/sageworks/web_components/model_markdown.py

@@ -30,7 +30,7 @@ def generate_markdown(self, model_details: dict) -> str:
         # Create simple markdown by iterating through the model_details dictionary
 
         # Excluded keys from the model_details dictionary (and any keys that end with '_arn')
-        exclude = ["size", "uuid"]
+        exclude = ["size", "uuid", "inference_meta"]
         top_level_details = {
             key: value for key, value in model_details.items() if key not in exclude and not key.endswith("_arn")
         }
@@ -51,14 +51,23 @@ def generate_markdown(self, model_details: dict) -> str:
             # Add to markdown string
             markdown += f"**{key}:** {value_str}  \n"
 
-        # Model Metrics
-        markdown += "### Model Metrics  \n"
-        if model_details["uuid"] == "abalone-regression":
-            markdown += "**Test Data:** Abalone_Regression_Test_2023_10_11  \n"
-            markdown += "**Test Data Hash:** ebea16fbc63574fe91dcac35a0b2432f  \n"
+        # Model Test Metrics
+        markdown += "### Model Test Metrics  \n"
+        if model_details.get("inference_meta"):
+            test_data = model_details["inference_meta"].get("test_data", " - ")
+            test_data_hash = model_details["inference_meta"].get("test_data_hash", " - ")
+            test_rows = model_details["inference_meta"].get("test_rows", " - ")
+            description = model_details["inference_meta"].get("description", " - ")
         else:
-            markdown += "**Test Data:** Wine_Classification_Test_2023_09_03  \n"
-            markdown += "**Test Data Hash:** cac35a0b2432febea16fbc63574fe91d  \n"
+            test_data = "AWS Training Capture"
+            test_data_hash = " N/A "
+            test_rows = " - "
+            description = " - "
+
+        markdown += f"**Test Data:** {test_data}  \n"
+        markdown += f"**Data Hash:** {test_data_hash}  \n"
+        markdown += f"**Test Rows:** {test_rows}  \n"
+        markdown += f"**Description:** {description}  \n"
 
         # Grab the Metrics from the model details
         metrics = model_details.get("model_metrics")