Fix shape inference to TF serving (#319)

(cherry picked from commit ec3a798)

Author: 1vn

pkg/workloads/cortex/tf_api/api.py

@@ -150,11 +150,17 @@ def create_raw_prediction_request(sample):
     prediction_request.model_spec.signature_name = signature_key
 
     for column_name, value in sample.items():
-        shape = [1]
         if util.is_list(value):
             shape = [len(value)]
+            for dim in signature_def[signature_key]["inputs"][column_name]["tensorShape"]["dim"][
+                1:
+            ]:
+                shape.append(int(dim["size"]))
+        else:
+            shape = [1]
+            value = [value]
         sig_type = signature_def[signature_key]["inputs"][column_name]["dtype"]
-        tensor_proto = tf.make_tensor_proto([value], dtype=DTYPE_TO_TF_TYPE[sig_type], shape=shape)
+        tensor_proto = tf.make_tensor_proto(value, dtype=DTYPE_TO_TF_TYPE[sig_type], shape=shape)
         prediction_request.inputs[column_name].CopyFrom(tensor_proto)
 
     return prediction_request