Removing the check for duplicate ids.

Change: 117280722

2 files changed, 2 insertions, 41 deletions

diff --git a/tensorflow/contrib/linear_optimizer/kernels/sdca_ops.cc b/tensorflow/contrib/linear_optimizer/kernels/sdca_ops.cc
index c2b403dbe0..852ae2a04f 100644
--- a/tensorflow/contrib/linear_optimizer/kernels/sdca_ops.cc
+++ b/tensorflow/contrib/linear_optimizer/kernels/sdca_ops.cc
@@ -510,20 +510,6 @@ class SdcaSolver : public OpKernel {
                     "The number of example ids (%ld) should match the number "
                     "of example weights (%d).",
                     example_ids.size(), num_examples)));
-    const int64 num_duplicate_example_ids = [&] {
-      // TODO(katsiapis): Benchmark and/or optimize.
-      std::unordered_set<StringPiece, StringPiece::Hasher> unique_ids(
-          example_ids.size());
-      for (size_t i = 0; i < example_ids.size(); ++i) {
-        unique_ids.emplace(example_ids(i));
-      }
-      return example_ids.size() - unique_ids.size();
-    }();
-    OP_REQUIRES(context, num_duplicate_example_ids == 0,
-                errors::InvalidArgument(strings::Printf(
-                    "Detected %lld duplicates in example_ids, which usually "
-                    "indicates a bug in the input data.",
-                    num_duplicate_example_ids)));
 
     OpMutableInputList sparse_weights_inputs;
     OP_REQUIRES_OK(context, context->mutable_input_list(
diff --git a/tensorflow/contrib/linear_optimizer/python/kernel_tests/sdca_ops_test.py b/tensorflow/contrib/linear_optimizer/python/kernel_tests/sdca_ops_test.py
index 3d812a2cbf..e973a88bb7 100644
--- a/tensorflow/contrib/linear_optimizer/python/kernel_tests/sdca_ops_test.py
+++ b/tensorflow/contrib/linear_optimizer/python/kernel_tests/sdca_ops_test.py
@@ -172,33 +172,6 @@ class SdcaOptimizerTest(TensorFlowTestCase):
           'No examples found or all examples have zero weight.'):
         lr.approximate_duality_gap().eval()
 
-  def testDuplicateExampleIds(self):
-    # Setup test data with 1 positive, and 1 negative example.
-    example_protos = [
-        make_example_proto(
-            {'age': [0],
-             'gender': [0]}, 0),
-        make_example_proto(
-            {'age': [1],
-             'gender': [1]}, 1),
-    ]
-    example_weights = [1.0, 1.0]
-    with self._single_threaded_test_session():
-      examples = make_example_dict(example_protos, example_weights)
-      examples['example_ids'] = ['duplicate_id'
-                                 for _ in examples['example_ids']]
-      variables = make_variable_dict(1, 1)
-      options = dict(symmetric_l2_regularization=0.5,
-                     symmetric_l1_regularization=0,
-                     loss_type='squared_loss')
-
-      lr = SdcaModel(CONTAINER, examples, variables, options)
-      tf.initialize_all_variables().run()
-      self.assertAllClose([0.0, 0.0], lr.predictions(examples).eval())
-      with self.assertRaisesOpError('Detected 1 duplicates in example_ids'):
-        lr.minimize().run()
-      self.assertAllClose([0.0, 0.0], lr.predictions(examples).eval())
-
 
 class SdcaWithLogisticLossTest(SdcaOptimizerTest):
   """SDCA optimizer test class for logistic loss."""
@@ -464,6 +437,8 @@ class SdcaWithLogisticLossTest(SdcaOptimizerTest):
                           rtol=1e-2,
                           atol=1e-2)
 
+  # TODO(katsiaspis): add a test for the case when examples at the end of an
+  # epoch are repeated, since example id may be duplicated.
 
 class SdcaWithLinearLossTest(SdcaOptimizerTest):
   """SDCA optimizer test class for linear (squared) loss."""