Implementation:Tensorflow Serving Classifier Test

Knowledge Sources	Tensorflow_Serving
Domains	Testing, Classification
Last Updated	2026-02-13 00:00 GMT

Overview

Test suite validating the TensorFlow Serving classifier servable, which handles classification inference requests.

Description

This test file provides comprehensive coverage of the TensorFlowClassifier implementation. It uses a FakeSession that simulates a TensorFlow session by parsing serialized Example protos from input tensors and producing classification classes and scores. The RegressorTest parameterized fixture (parameterized on method name checking) validates classification with example lists, context, named signatures, error handling for malformed inputs, and correctness of tensor shapes and types.

Key areas tested include:

• Classification with example lists and example lists with context
• Classes-only and scores-only classification outputs
• Named and invalid signature handling
• Error cases: empty inputs, malformed scores, incorrect tensor batch sizes, incorrect tensor types, mismatched class counts
• Method name checking feature toggle

Usage

Run these tests to validate changes to the classifier servable, signature handling, or input/output tensor processing for classification tasks.

Code Reference

Source Location

• Repository: Tensorflow_Serving
• File: tensorflow_serving/servables/tensorflow/classifier_test.cc
• Lines: 1-1018

Test Fixture

class FakeSession : public tensorflow::Session {
 public:
  explicit FakeSession(absl::optional<int64_t> expected_timeout)
      : expected_timeout_(expected_timeout) {}
  ~FakeSession() override = default;
  // Parses serialized Examples from input tensors and produces
  // classification classes and scores as output tensors.
  absl::Status Run(const RunOptions& run_options,
                   const std::vector<std::pair<string, Tensor>>& inputs,
                   const std::vector<string>& output_names,
                   const std::vector<string>& target_nodes,
                   std::vector<Tensor>* outputs,
                   RunMetadata* run_metadata,
                   const thread::ThreadPoolOptions& thread_pool_options) override;
};

class ClassifierTest : public ::testing::TestWithParam<bool> {
  // Parameterized on method name check enabled/disabled
};

Build Target

bazel test //tensorflow_serving/servables/tensorflow:classifier_test

Test Coverage

Key Test Cases

Test Name	Category	Description
ExampleList	Input Handling	Tests classification with a standard example list
ExampleListWithContext	Input Handling	Tests classification with example list and context features
ExampleListWithContext_DuplicateFeatures	Input Handling	Tests context feature merging with duplicate features
ClassesOnly	Output Handling	Tests classification returning only class labels
ScoresOnly	Output Handling	Tests classification returning only scores
ZeroScoresArePresent	Output Handling	Verifies zero-valued scores are correctly included
ValidNamedSignature	Signature	Tests classification with a valid named signature
InvalidNamedSignature	Error Handling	Tests error on invalid named signature
MalformedScores	Error Handling	Tests error on improperly sized score tensors
MissingClassificationSignature	Error Handling	Tests error when classification signature is missing
EmptyInput	Error Handling	Tests error on empty input
EmptyExampleList	Error Handling	Tests error on empty example list
EmptyExampleListWithContext	Error Handling	Tests error on empty example list with context
RunsFails	Error Handling	Tests handling of session Run failures
ClassesIncorrectTensorBatchSize	Validation	Tests error on mismatched class tensor batch size
ClassesIncorrectTensorType	Validation	Tests error on incorrect class tensor type
ScoresIncorrectTensorBatchSize	Validation	Tests error on mismatched score tensor batch size
ScoresIncorrectTensorType	Validation	Tests error on incorrect score tensor type
MismatchedNumberOfTensorClasses	Validation	Tests error when class and score tensor sizes differ
MethodNameCheck	Feature Toggle	Tests method name checking feature

Usage Examples

Test Pattern

TEST_P(ClassifierTest, ExampleList) {
  auto* examples =
      request_.mutable_input()->mutable_example_list()->mutable_examples();
  *examples->Add() = example_with_class({{"car", "train", "bicycle"}});
  *examples->Add() = example_with_class({{"car", "train", "bicycle"}});

  ClassificationResult result;
  TF_ASSERT_OK(classifier_->Classify(request_, &result));
  EXPECT_THAT(result, EqualsProto(expected_result));
}

Related Pages

• Implementation:Tensorflow_Serving_Tfrt_Classifier_Test