Update python tests

examples/insurance/implementations/models/dnn.py

@@ -2,27 +2,33 @@
 
 
 def create_estimator(run_config, model_config):
+    aggregates = model_config["input"]["aggregates"]
+
     feature_columns = [
         tf.feature_column.indicator_column(
-            tf.feature_column.categorical_column_with_vocabulary_list("sex", ["female", "male"])
+            tf.feature_column.categorical_column_with_vocabulary_list(
+                "sex", aggregates["sex_vocab"]
+            )
         ),
         tf.feature_column.indicator_column(
-            tf.feature_column.categorical_column_with_vocabulary_list("smoker", ["yes", "no"])
+            tf.feature_column.categorical_column_with_vocabulary_list(
+                "smoker", aggregates["smoker_vocab"]
+            )
         ),
         tf.feature_column.indicator_column(
             tf.feature_column.categorical_column_with_vocabulary_list(
-                "region", ["northwest", "northeast", "southwest", "southeast"]
+                "region", aggregates["region_vocab"]
             )
         ),
         tf.feature_column.bucketized_column(
-            tf.feature_column.numeric_column("age"), [15, 20, 25, 35, 40, 45, 50, 55, 60, 65]
+            tf.feature_column.numeric_column("age"), aggregates["age_buckets"]
         ),
         tf.feature_column.bucketized_column(
-            tf.feature_column.numeric_column("bmi"), [15, 20, 25, 35, 40, 45, 50, 55]
+            tf.feature_column.numeric_column("bmi"), aggregates["bmi_buckets"]
         ),
         tf.feature_column.indicator_column(
             tf.feature_column.categorical_column_with_vocabulary_list(
-                "children", model_config["aggregates"]["children_set"]
+                "children", aggregates["children_set"]
             )
         ),
     ]

examples/insurance/resources/apis.yaml

@@ -1,5 +1,5 @@
 - kind: api
   name: cost
-  model_name: dnn
+  model: @dnn
   compute:
     replicas: 1

examples/insurance/resources/environments.yaml

@@ -3,11 +3,4 @@
   data:
     type: csv
     path: s3a://cortex-examples/insurance.csv
-    schema:
-      - age
-      - sex
-      - bmi
-      - children
-      - smoker
-      - region
-      - charges
+    schema: [@age, @sex, @bmi, @children, @smoker, @region, @charges]

examples/insurance/resources/features.yaml

@@ -47,30 +47,22 @@
 - kind: aggregate
   name: charges_mean
   aggregator: cortex.mean
-  inputs:
-    columns:
-      col: charges
+  input: @charges
 
 - kind: aggregate
   name: charges_stddev
   aggregator: cortex.stddev
-  inputs:
-    columns:
-      col: charges
+  input: @charges
 
 - kind: aggregate
   name: children_set
   aggregator: cortex.collect_set_int
-  inputs:
-    columns:
-      col: children
+  input: @children
 
 - kind: transformed_column
   name: charges_normalized
   transformer: cortex.normalize
-  inputs:
-    columns:
-      num: charges
-    args:
-      mean: charges_mean
-      stddev: charges_stddev
+  input:
+    col: @charges
+    mean: @charges_mean
+    stddev: @charges_stddev

examples/insurance/resources/models.yaml

@@ -1,20 +1,20 @@
 - kind: model
   name: dnn
-  type: regression
-  target_column: charges_normalized
-  feature_columns:
-    - age
-    - sex
-    - bmi
-    - children
-    - smoker
-    - region
+  estimator_path: implementations/models/dnn.py
+  target_column: @charges_normalized
+  input:
+    features: [@age, @sex, @bmi, @children, @smoker, @region, @charges]
+    aggregates:
+      children_set: @children_set
+      region_vocab: ["northwest", "northeast", "southwest", "southeast"]
+      age_buckets: [15, 20, 25, 35, 40, 45, 50, 55, 60, 65]
+      bmi_buckets: [15, 20, 25, 35, 40, 45, 50, 55]
+      smoker_vocab: ["yes", "no"]
+      sex_vocab: ["female", "male"]
   hparams:
     hidden_units: [100, 100, 100]
   data_partition_ratio:
     training: 0.8
     evaluation: 0.2
   training:
     num_steps: 10000
-  aggregates:
-    - children_set

pkg/workloads/lib/context.py

@@ -721,8 +721,8 @@ def _deserialize_raw_ctx(raw_ctx):
 # input should already have non-column arguments replaced, and all types validated
 def create_transformer_inputs_from_map(input, col_value_map):
     if util.is_str(input):
-        res_name = util.get_resource_ref(input)
-        if res_name is not None and res_name in col_value_map:
+        if util.is_resource_ref(input):
+            res_name = util.get_resource_ref(input)
             return col_value_map[res_name]
         return input
 

pkg/workloads/lib/util.py

@@ -879,19 +879,18 @@ def is_resource_ref(obj):
 
 def get_resource_ref(obj):
     if not is_str(obj):
-        return None
+        raise ValueError("expected input of type string but received " + str(type(obj)))
     if obj.startswith(resource_escape_seq):
         return obj[len(resource_escape_seq) :]
     elif obj.startswith(resource_escape_seq_raw):
         return obj[len(resource_escape_seq_raw) :]
-    return None
+    raise ValueError("expected a resource reference but got " + obj)
 
 
 def extract_resource_refs(input):
     if is_str(input):
-        res = get_resource_ref(input)
-        if res is not None:
-            return {res}
+        if is_resource_ref(input):
+            return {get_resource_ref(input)}
         return set()
 
     if is_list(input):