TensorFlow: Upstream changes from since last Thursday.

Changes:
- Bug fix to input.py by @dave-andersen with tests.

- Some fixes to names in the whitepaper

- Include cfloat for FLT_MAX by @benoitsteiner

- Fix broken link in mnist tutorials by @mrry

- Typos and fixes to documentation by @Sohl-Dickstein, @ebrevdo, @vrv,
  Yaroslav, @martinwicke

- Fixed unnecessary check around a delete by @girving

- More compatibility tests by @josh11b

- Misc other typos and fixes by Googlers.

Base CL: 107944070

35 files changed, 488 insertions, 78 deletions

diff --git a/tensorflow/core/framework/op_compatibility_test.cc b/tensorflow/core/framework/op_compatibility_test.cc
index 7716b34ca1..0f5e27de79 100644
--- a/tensorflow/core/framework/op_compatibility_test.cc
+++ b/tensorflow/core/framework/op_compatibility_test.cc
@@ -49,6 +49,7 @@ class OpCompatibilityTest : public OpsTestBase {
                                 &new_out_types));
     ASSERT_EQ(new_in_types, old_in_types);
     ASSERT_EQ(new_out_types, old_out_types);
+    ASSERT_OK(OpDefCompatible(old_op_def, *new_op_def));
 
     // Verify the Op actually runs.  Result() will return the output.
     ASSERT_OK(InitOp());
@@ -57,7 +58,21 @@ class OpCompatibilityTest : public OpsTestBase {
 
   string Result() { return GetOutput(0)->scalar<string>()(); }
 
-  void ExpectInvalid(const OpDef& old_op_def, string error) {
+  void ExpectIncompatible(const OpDef& old_op_def, const OpDef& new_op_def,
+                          const string& error) {
+    // Test OpDefCompatible gives the same answer without the node_def.
+    Status status = OpDefCompatible(old_op_def, new_op_def);
+    if (status.ok()) {
+      ADD_FAILURE() << SummarizeOpDef(old_op_def) << " vs. "
+                    << SummarizeOpDef(new_op_def);
+    } else {
+      EXPECT_TRUE(StringPiece(status.error_message()).contains(error))
+          << status << " does not contain " << error;
+    }
+  }
+
+  void ExpectInvalid(const OpDef& old_op_def, const string& validation_error,
+                     const string& compatibility_error) {
     // Record the original signature before we change *node_def().
     DataTypeVector old_in_types, old_out_types;
     ASSERT_OK(InOutTypesForNode(*node_def(), old_op_def, &old_in_types,
@@ -72,12 +87,16 @@ class OpCompatibilityTest : public OpsTestBase {
     if (status.ok()) {
       ADD_FAILURE() << SummarizeNodeDef(*node_def());
     } else {
-      EXPECT_TRUE(StringPiece(status.error_message()).contains(error))
-          << status << " does not contain " << error;
+      EXPECT_TRUE(
+          StringPiece(status.error_message()).contains(validation_error))
+          << status << " does not contain " << validation_error;
     }
+
+    ExpectIncompatible(old_op_def, *new_op_def, compatibility_error);
   }
 
-  void ExpectTypeMismatch(const OpDef& old_op_def) {
+  void ExpectTypeMismatch(const OpDef& old_op_def,
+                          const string& compatibility_error) {
     // Record the original signature before we change *node_def().
     DataTypeVector old_in_types, old_out_types;
     ASSERT_OK(InOutTypesForNode(*node_def(), old_op_def, &old_in_types,
@@ -98,6 +117,8 @@ class OpCompatibilityTest : public OpsTestBase {
                     << DataTypeVectorString(new_in_types) << " -> "
                     << DataTypeVectorString(new_out_types);
     }
+
+    ExpectIncompatible(old_op_def, *new_op_def, compatibility_error);
   }
 };
 
@@ -338,6 +359,57 @@ TEST_F(OpCompatibilityTest, PolyIntoList) {
   EXPECT_EQ("poly_into_list = PolyIntoList[N=1, T=DT_INT32](a)", Result());
 }
 
+// Should be able to make a multiple inputs/outputs into a list with
+// the default types matching the inputs/outputs being replaced.
+
+// Changing from int32, int32 -> N * int32 (where N: int = 2 by default).
+REGISTER_OP("MakeMultipleSameList")
+    .Input("a: N * int32")
+    .Output("ndef: string")
+    .Attr("N: int = 2");
+REGISTER_KERNEL_BUILDER(Name("MakeMultipleSameList").Device(DEVICE_CPU),
+                        TestKernel);
+
+TEST_F(OpCompatibilityTest, MakeMultipleSameList) {
+  OpDef old_op_def;
+  ASSERT_OK(OpDefBuilder("MakeMultipleSameList")
+                .Input("a: int32")
+                .Input("b: int32")
+                .Output("ndef: string")
+                .Finalize(&old_op_def));
+  ASSERT_OK(NodeDefBuilder("make_list", &old_op_def)
+                .Input(FakeInput())
+                .Input(FakeInput())
+                .Finalize(node_def()));
+  ExpectSuccess(old_op_def);
+  EXPECT_EQ("make_list = MakeMultipleSameList[N=2](a, b)", Result());
+}
+
+// Changing from int32, float -> T
+// (where T: list(type) = [int32, float] by default).
+REGISTER_OP("MakeMultipleAnyList")
+    .Input("a: T")
+    .Output("ndef: string")
+    .Attr("T: list(type) = [DT_INT32, DT_FLOAT]");
+REGISTER_KERNEL_BUILDER(Name("MakeMultipleAnyList").Device(DEVICE_CPU),
+                        TestKernel);
+
+TEST_F(OpCompatibilityTest, MakeMultipleAnyList) {
+  OpDef old_op_def;
+  ASSERT_OK(OpDefBuilder("MakeMultipleAnyList")
+                .Input("a: int32")
+                .Input("b: float")
+                .Output("ndef: string")
+                .Finalize(&old_op_def));
+  ASSERT_OK(NodeDefBuilder("make_list", &old_op_def)
+                .Input(FakeInput())
+                .Input(FakeInput())
+                .Finalize(node_def()));
+  ExpectSuccess(old_op_def);
+  EXPECT_EQ("make_list = MakeMultipleAnyList[T=[DT_INT32, DT_FLOAT]](a, b)",
+            Result());
+}
+
 // Should be able to change the name of an input/output.
 REGISTER_OP("ChangeName").Input("y: int32").Output("ndef: string");
 REGISTER_KERNEL_BUILDER(Name("ChangeName").Device(DEVICE_CPU), TestKernel);
@@ -477,6 +549,18 @@ TEST_F(OpCompatibilityTest, ShorterSameList) {
 
 // Negative tests -------------------------------------------------------------
 
+// Can't remove an attr.
+REGISTER_OP("RemoveAttr");
+
+TEST_F(OpCompatibilityTest, RemoveAttrFails) {
+  OpDef old_op_def;
+  ASSERT_OK(OpDefBuilder("RemoveAttr").Attr("a: int").Finalize(&old_op_def));
+  ASSERT_OK(
+      NodeDefBuilder("fails", &old_op_def).Attr("a", 3).Finalize(node_def()));
+  ExpectInvalid(old_op_def, "NodeDef mentions attr 'a' not in",
+                "Attr 'a' removed");
+}
+
 // Can't add an attr without a default.
 REGISTER_OP("AddAttrNoDefault").Attr("a: int");
 
@@ -484,7 +568,8 @@ TEST_F(OpCompatibilityTest, AddAttrNoDefaultFails) {
   OpDef old_op_def;
   ASSERT_OK(OpDefBuilder("AddAttrNoDefault").Finalize(&old_op_def));
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def).Finalize(node_def()));
-  ExpectInvalid(old_op_def, "NodeDef missing attr 'a'");
+  ExpectInvalid(old_op_def, "NodeDef missing attr 'a'",
+                "Attr 'a' added without default");
 }
 
 // Can't add a non-list input/output.
@@ -495,7 +580,8 @@ TEST_F(OpCompatibilityTest, AddSingleInputFails) {
   ASSERT_OK(OpDefBuilder("AddSingleInput").Finalize(&old_op_def));
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def).Finalize(node_def()));
   ExpectInvalid(old_op_def,
-                "expected inputs 'int32' do not match 0 inputs specified");
+                "expected inputs 'int32' do not match 0 inputs specified",
+                "Input signature mismatch '' vs. 'int32'");
 }
 
 // Can't add a list input/output without an empty default.
@@ -514,7 +600,8 @@ TEST_F(OpCompatibilityTest, AddNIntsBigDefaultFails) {
   ASSERT_OK(OpDefBuilder("AddNIntsBigDefault").Finalize(&old_op_def));
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def).Finalize(node_def()));
   ExpectInvalid(old_op_def,
-                "expected inputs 'int32' do not match 0 inputs specified");
+                "expected inputs 'int32' do not match 0 inputs specified",
+                "Input signature mismatch '' vs. 'int32'");
 }
 
 TEST_F(OpCompatibilityTest, AddNSameBigDefaultFails) {
@@ -522,7 +609,8 @@ TEST_F(OpCompatibilityTest, AddNSameBigDefaultFails) {
   ASSERT_OK(OpDefBuilder("AddNSameBigDefault").Finalize(&old_op_def));
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def).Finalize(node_def()));
   ExpectInvalid(old_op_def,
-                "expected inputs 'int32' do not match 0 inputs specified");
+                "expected inputs 'int32' do not match 0 inputs specified",
+                "Input signature mismatch '' vs. 'int32'");
 }
 
 TEST_F(OpCompatibilityTest, AddListBigDefaultFails) {
@@ -530,7 +618,8 @@ TEST_F(OpCompatibilityTest, AddListBigDefaultFails) {
   ASSERT_OK(OpDefBuilder("AddListBigDefault").Finalize(&old_op_def));
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def).Finalize(node_def()));
   ExpectInvalid(old_op_def,
-                "expected inputs 'int32' do not match 0 inputs specified");
+                "expected inputs 'int32' do not match 0 inputs specified",
+                "Input signature mismatch '' vs. 'int32'");
 }
 
 // Can't change the type of an input/output.
@@ -543,7 +632,8 @@ TEST_F(OpCompatibilityTest, ChangeTypeFails) {
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def)
                 .Input(FakeInput())
                 .Finalize(node_def()));
-  ExpectTypeMismatch(old_op_def);
+  ExpectTypeMismatch(old_op_def,
+                     "Input signature mismatch 'int32' vs. 'float'");
 }
 
 // Can't change the order of inputs/outputs.
@@ -560,7 +650,24 @@ TEST_F(OpCompatibilityTest, ChangeOrderFails) {
                 .Input(FakeInput())
                 .Input(FakeInput())
                 .Finalize(node_def()));
-  ExpectTypeMismatch(old_op_def);
+  ExpectTypeMismatch(
+      old_op_def, "Input signature mismatch 'int32, float' vs. 'float, int32'");
+}
+
+// Can't remove inputs/outputs.
+
+REGISTER_OP("RemoveInput");
+
+TEST_F(OpCompatibilityTest, RemoveInputFails) {
+  OpDef old_op_def;
+  ASSERT_OK(
+      OpDefBuilder("RemoveInput").Input("a: float").Finalize(&old_op_def));
+  ASSERT_OK(NodeDefBuilder("fails", &old_op_def)
+                .Input(FakeInput())
+                .Finalize(node_def()));
+  ExpectInvalid(old_op_def,
+                "expected inputs '' do not match 1 inputs specified",
+                "Input signature mismatch 'float' vs. ''");
 }
 
 // Can't change the type of an attr.
@@ -574,7 +681,8 @@ TEST_F(OpCompatibilityTest, ChangeAttrTypeFails) {
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def)
                 .Attr("a", true)
                 .Finalize(node_def()));
-  ExpectInvalid(old_op_def, "value with type 'bool' when 'int' expected");
+  ExpectInvalid(old_op_def, "value with type 'bool' when 'int' expected",
+                "Attr 'a' changed type 'bool' -> 'int'");
 }
 
 // Can't change an attr from a list.
@@ -587,7 +695,8 @@ TEST_F(OpCompatibilityTest, AttrFromListFails) {
       OpDefBuilder("AttrFromList").Attr("a: list(int)").Finalize(&old_op_def));
   ASSERT_OK(
       NodeDefBuilder("fails", &old_op_def).Attr("a", {5}).Finalize(node_def()));
-  ExpectInvalid(old_op_def, "value with type 'list(int)' when 'int' expected");
+  ExpectInvalid(old_op_def, "value with type 'list(int)' when 'int' expected",
+                "Attr 'a' changed type 'list(int)' -> 'int'");
 }
 
 // Can't change an attr to a list.
@@ -599,7 +708,8 @@ TEST_F(OpCompatibilityTest, AttrToListFails) {
   ASSERT_OK(OpDefBuilder("AttrToList").Attr("a: int").Finalize(&old_op_def));
   ASSERT_OK(
       NodeDefBuilder("fails", &old_op_def).Attr("a", 5).Finalize(node_def()));
-  ExpectInvalid(old_op_def, "value with type 'int' when 'list(int)' expected");
+  ExpectInvalid(old_op_def, "value with type 'int' when 'list(int)' expected",
+                "Attr 'a' changed type 'int' -> 'list(int)'");
 }
 
 // Can't change an input from polymorphic to a list of any type.
@@ -615,8 +725,8 @@ TEST_F(OpCompatibilityTest, PolymorphicToAnyListFails) {
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def)
                 .Input(FakeInput(DT_INT32))
                 .Finalize(node_def()));
-  ExpectInvalid(old_op_def,
-                "value with type 'type' when 'list(type)' expected");
+  ExpectInvalid(old_op_def, "value with type 'type' when 'list(type)' expected",
+                "Attr 'T' changed type 'type' -> 'list(type)'");
 }
 
 // Can't change an input from a list of the same type to a list of any type.
@@ -636,8 +746,8 @@ TEST_F(OpCompatibilityTest, SameToAnyListFails) {
   ASSERT_OK(NodeDefBuilder("fails", &old_op_def)
                 .Input(FakeInput(1, DT_INT32))
                 .Finalize(node_def()));
-  ExpectInvalid(old_op_def,
-                "value with type 'type' when 'list(type)' expected");
+  ExpectInvalid(old_op_def, "value with type 'type' when 'list(type)' expected",
+                "Attr 'T' changed type 'type' -> 'list(type)'");
 }
 
 // Changing an attr's default is not technically illegal, but should
diff --git a/tensorflow/core/framework/op_def_util.cc b/tensorflow/core/framework/op_def_util.cc
index e3aef011de..7b0fa668bf 100644
--- a/tensorflow/core/framework/op_def_util.cc
+++ b/tensorflow/core/framework/op_def_util.cc
@@ -1,13 +1,14 @@
 #include "tensorflow/core/framework/op_def_util.h"
 
 #include <set>
+#include <unordered_map>
 #include "tensorflow/core/framework/attr_value_util.h"
-#include "tensorflow/core/platform/protobuf.h"
 #include "tensorflow/core/framework/types.h"
+#include "tensorflow/core/lib/core/errors.h"
 #include "tensorflow/core/lib/core/stringpiece.h"
 #include "tensorflow/core/lib/gtl/map_util.h"
-#include "tensorflow/core/lib/core/errors.h"
 #include "tensorflow/core/platform/port.h"
+#include "tensorflow/core/platform/protobuf.h"
 #include "tensorflow/core/platform/regexp.h"
 
 namespace tensorflow {
@@ -341,4 +342,161 @@ string SummarizeOpDef(const OpDef& op_def) {
   return ret;
 }
 
+namespace {
+
+typedef std::unordered_map<string, const OpDef::AttrDef*> AttrMap;
+void FillAttrMap(const OpDef& op_def, AttrMap* attr_map) {
+  for (const auto& attr : op_def.attr()) {
+    (*attr_map)[attr.name()] = &attr;
+  }
+}
+
+// Add a comma to *s every call but the first (*add_comma should be
+// initialized to false).
+void AddComma(string* s, bool* add_comma) {
+  if (*add_comma) {
+    strings::StrAppend(s, ", ");
+  } else {
+    *add_comma = true;
+  }
+}
+
+// Compute a signature for either inputs or outputs that will be the
+// same for both the old and new OpDef if they are compatible.  We
+// assume that new_attrs is a superset of old_attrs, and that any attr
+// in the difference has a default.  Our strategy is to make a list of
+// types, where the types are things like:
+// * "int32", "float", etc.,
+// * "T" for some attr "T" in old_attrs, or
+// * "N * type" for "N" either some attr in old_attrs.
+//
+// We get the types by either using the attrs in args if they are in
+// old_attrs, or substituting the default value from new_attrs.
+string ComputeArgSignature(
+    const protobuf::RepeatedPtrField<OpDef::ArgDef>& args,
+    const AttrMap& old_attrs, const AttrMap& new_attrs) {
+  string s;
+  bool add_comma = false;
+  for (const OpDef::ArgDef& arg : args) {
+    if (!arg.type_list_attr().empty()) {
+      const OpDef::AttrDef* old_attr =
+          gtl::FindPtrOrNull(old_attrs, arg.type_list_attr());
+      if (old_attr) {
+        // Both old and new have the list(type) attr, so can use it directly.
+        AddComma(&s, &add_comma);
+        strings::StrAppend(&s, arg.type_list_attr());
+        if (arg.is_ref()) strings::StrAppend(&s, " ref");
+      } else {
+        // Missing the list(type) attr in the old, so use the default
+        // value for the attr from new instead.
+        const OpDef::AttrDef* new_attr =
+            gtl::FindPtrOrNull(new_attrs, arg.type_list_attr());
+        const auto& type_list = new_attr->default_value().list().type();
+        if (type_list.empty()) continue;
+        for (int i = 0; i < type_list.size(); ++i) {
+          AddComma(&s, &add_comma);
+          strings::StrAppend(
+              &s, DataTypeString(static_cast<DataType>(type_list.Get(i))));
+          if (arg.is_ref()) strings::StrAppend(&s, " ref");
+        }
+      }
+    } else {
+      int num = 1;  // How many input/outputs does this represent?
+      if (!arg.number_attr().empty()) {
+        // N * type case.
+        const OpDef::AttrDef* old_attr =
+            gtl::FindPtrOrNull(old_attrs, arg.number_attr());
+        if (old_attr) {
+          // Both old and new have the number attr, so can use it directly.
+          AddComma(&s, &add_comma);
+          strings::StrAppend(&s, arg.number_attr(), " * ");
+          add_comma = false;  // Don't add another comma before the type.
+        } else {
+          // Missing the number attr in the old, so use the default
+          // value for the attr from new instead.
+          const OpDef::AttrDef* new_attr =
+              gtl::FindPtrOrNull(new_attrs, arg.number_attr());
+          num = new_attr->default_value().i();
+        }
+      }
+
+      string type;  // What is the type of this arg?
+      if (arg.type() != DT_INVALID) {
+        // int32, float, etc. case
+        type = DataTypeString(arg.type());
+      } else {
+        const OpDef::AttrDef* old_attr =
+            gtl::FindPtrOrNull(old_attrs, arg.type_attr());
+        if (old_attr) {
+          // Both old and new have the type attr, so can use it directly.
+          type = arg.type_attr();
+        } else {
+          // Missing the type attr in the old, so use the default
+          // value for the attr from new instead.
+          const OpDef::AttrDef* new_attr =
+              gtl::FindPtrOrNull(new_attrs, arg.type_attr());
+          type = DataTypeString(new_attr->default_value().type());
+        }
+      }
+      if (arg.is_ref()) strings::StrAppend(&type, " ref");
+
+      // Record `num` * `type` in the signature.
+      for (int i = 0; i < num; ++i) {
+        AddComma(&s, &add_comma);
+        strings::StrAppend(&s, type);
+      }
+    }
+  }
+
+  return s;
+}
+
+}  // namespace
+
+Status OpDefCompatible(const OpDef& old_op, const OpDef& new_op) {
+#define VALIDATE(CONDITION, ...)                                            \
+  if (!(CONDITION)) {                                                       \
+    return errors::InvalidArgument("Incompatible Op change: ", __VA_ARGS__, \
+                                   "; old: ", SummarizeOpDef(old_op),       \
+                                   "; new: ", SummarizeOpDef(new_op));      \
+  }
+
+  VALIDATE(old_op.name() == new_op.name(), "Name mismatch");
+
+  AttrMap new_attrs, old_attrs;
+  FillAttrMap(old_op, &old_attrs);
+  FillAttrMap(new_op, &new_attrs);
+  for (const auto& old_attr : old_op.attr()) {
+    const OpDef::AttrDef* new_attr =
+        gtl::FindPtrOrNull(new_attrs, old_attr.name());
+    VALIDATE(new_attr != nullptr, "Attr '", old_attr.name(), "' removed");
+    VALIDATE(old_attr.type() == new_attr->type(), "Attr '", old_attr.name(),
+             "' changed type '", old_attr.type(), "' -> '", new_attr->type(),
+             "'");
+  }
+
+  for (const auto& new_attr : new_op.attr()) {
+    const OpDef::AttrDef* old_attr =
+        gtl::FindPtrOrNull(old_attrs, new_attr.name());
+    VALIDATE(old_attr != nullptr || new_attr.has_default_value(), "Attr '",
+             new_attr.name(), "' added without default");
+  }
+
+  const string old_in_sig =
+      ComputeArgSignature(old_op.input_arg(), old_attrs, new_attrs);
+  const string new_in_sig =
+      ComputeArgSignature(new_op.input_arg(), old_attrs, new_attrs);
+  VALIDATE(old_in_sig == new_in_sig, "Input signature mismatch '", old_in_sig,
+           "' vs. '", new_in_sig, "'");
+
+  const string old_out_sig =
+      ComputeArgSignature(old_op.output_arg(), old_attrs, new_attrs);
+  const string new_out_sig =
+      ComputeArgSignature(new_op.output_arg(), old_attrs, new_attrs);
+  VALIDATE(old_out_sig == new_out_sig, "Output signature mismatch '",
+           old_out_sig, "' vs. '", new_out_sig, "'");
+
+  return Status::OK();
+}
+
 }  // namespace tensorflow
diff --git a/tensorflow/core/framework/op_def_util.h b/tensorflow/core/framework/op_def_util.h
index a9fecf3fa0..c8440c5def 100644
--- a/tensorflow/core/framework/op_def_util.h
+++ b/tensorflow/core/framework/op_def_util.h
@@ -27,6 +27,11 @@ OpDef::AttrDef* FindAttrMutable(StringPiece name, OpDef* op_def);
 // than a text-format proto.  Excludes descriptions.
 string SummarizeOpDef(const OpDef& op_def);
 
+// Returns an error if new_op is not backwards-compatible with (more
+// accepting than) old_op.
+// REQUIRES: old_op and new_op must pass validation.
+Status OpDefCompatible(const OpDef& old_op, const OpDef& new_op);
+
 }  // namespace tensorflow
 
 #endif  // TENSORFLOW_FRAMEWORK_OP_DEF_UTIL_H_
diff --git a/tensorflow/core/kernels/maxpooling_op_gpu.cu.cc b/tensorflow/core/kernels/maxpooling_op_gpu.cu.cc
index 65262eb54e..13684c82f7 100644
--- a/tensorflow/core/kernels/maxpooling_op_gpu.cu.cc
+++ b/tensorflow/core/kernels/maxpooling_op_gpu.cu.cc
@@ -3,6 +3,7 @@
 #define EIGEN_USE_GPU
 
 #include <stdio.h>
+#include <cfloat>
 
 #include "tensorflow/core/framework/register_types.h"
 #include "tensorflow/core/framework/tensor_types.h"
diff --git a/tensorflow/core/kernels/reverse_op.h b/tensorflow/core/kernels/reverse_op.h
index bba25f70e8..ca13b98a71 100644
--- a/tensorflow/core/kernels/reverse_op.h
+++ b/tensorflow/core/kernels/reverse_op.h
@@ -22,6 +22,16 @@ struct Reverse {
   }
 };
 
+template <typename Device, typename T>
+struct Reverse<Device, T, 0> {
+  void operator()(const Device& d, typename TTypes<T, 0>::ConstTensor input,
+                  typename TTypes<bool, 1>::ConstTensor,
+                  typename TTypes<T, 0>::Tensor output) {
+    // Reversing a scalar is copying it.
+    output.device(d) = input;
+  }
+};
+
 }  // namespace functor
 }  // namespace tensorflow
 
diff --git a/tensorflow/core/util/tensor_slice_reader_cache.cc b/tensorflow/core/util/tensor_slice_reader_cache.cc
index af81d0115e..24eedd5c5c 100644
--- a/tensorflow/core/util/tensor_slice_reader_cache.cc
+++ b/tensorflow/core/util/tensor_slice_reader_cache.cc
@@ -9,9 +9,7 @@ namespace checkpoint {
 
 TensorSliceReaderCacheWrapper::TensorSliceReaderCacheWrapper() {}
 TensorSliceReaderCacheWrapper::~TensorSliceReaderCacheWrapper() {
-  if (cache_) {
-    delete cache_;
-  }
+  delete cache_;
   cache_ = nullptr;
 }
 
diff --git a/tensorflow/g3doc/api_docs/leftnav_files b/tensorflow/g3doc/api_docs/leftnav_files
new file mode 100644
index 0000000000..ceeb88af8f
--- /dev/null
+++ b/tensorflow/g3doc/api_docs/leftnav_files
@@ -0,0 +1,30 @@
+### [Overview](/api_docs/index.md)
+### [Python API](/api_docs/python/index.md)
+python/framework.md
+python/constant_op.md
+python/state_ops.md
+python/array_ops.md
+python/math_ops.md
+python/control_flow_ops.md
+python/image.md
+python/sparse_ops.md
+python/io_ops.md
+python/python_io.md
+python/nn.md
+python/client.md
+python/train.md
+### [C++ API](/api_docs/cc/index.md)
+cc/ClassEnv.md
+cc/ClassRandomAccessFile.md
+cc/ClassWritableFile.md
+cc/ClassEnvWrapper.md
+cc/ClassSession.md
+cc/StructSessionOptions.md
+cc/ClassStatus.md
+cc/StructState.md
+cc/ClassTensor.md
+cc/ClassTensorShape.md
+cc/StructTensorShapeDim.md
+cc/ClassTensorShapeUtils.md
+cc/ClassThread.md
+cc/StructThreadOptions.md
\ No newline at end of file
diff --git a/tensorflow/g3doc/api_docs/python/array_ops.md b/tensorflow/g3doc/api_docs/python/array_ops.md
index 10a6df41ff..1b05a48862 100644
--- a/tensorflow/g3doc/api_docs/python/array_ops.md
+++ b/tensorflow/g3doc/api_docs/python/array_ops.md
@@ -329,7 +329,7 @@ reshape(t, [3, 3]) ==> [[1, 2, 3]
 
 # tensor 't' is [[[1, 1], [2, 2]]
 #                [[3, 3], [4, 4]]]
-# tensor 't' has shape [2, 2]
+# tensor 't' has shape [2, 2, 2]
 reshape(t, [2, 4]) ==> [[1, 1, 2, 2]
                         [3, 3, 4, 4]]
 
diff --git a/tensorflow/g3doc/api_docs/python/framework.md b/tensorflow/g3doc/api_docs/python/framework.md
index 1f88ad0ca4..eab4ec0152 100644
--- a/tensorflow/g3doc/api_docs/python/framework.md
+++ b/tensorflow/g3doc/api_docs/python/framework.md
@@ -147,6 +147,11 @@ This method is thread-safe.
   A [`GraphDef`](https://tensorflow.googlesource.com/tensorflow/+/master/tensorflow/core/framework/graph.proto)
   protocol buffer.
 
+##### Raises: <a class="md-anchor" id="AUTOGENERATED-raises-"></a>
+
+
+*  <b>`ValueError`</b>: If the graph_def would be too large.
+
 
 - - -
 
diff --git a/tensorflow/g3doc/api_docs/python/io_ops.md b/tensorflow/g3doc/api_docs/python/io_ops.md
index a728ba579e..b10cc79bc8 100644
--- a/tensorflow/g3doc/api_docs/python/io_ops.md
+++ b/tensorflow/g3doc/api_docs/python/io_ops.md
@@ -65,7 +65,7 @@ be fed using the `feed_dict` optional argument to `Session.run()`,
 For example:
 
 ```python
-x = tf.placeholder(float, shape=(1024, 1024))
+x = tf.placeholder(tf.float32, shape=(1024, 1024))
 y = tf.matmul(x, x)
 
 with tf.Session() as sess:
diff --git a/tensorflow/g3doc/api_docs/python/nn.md b/tensorflow/g3doc/api_docs/python/nn.md
index 4f4292ff5e..8cb1f96e09 100644
--- a/tensorflow/g3doc/api_docs/python/nn.md
+++ b/tensorflow/g3doc/api_docs/python/nn.md
@@ -139,7 +139,8 @@ kept independently and each row and column will be kept or not kept together.
 
 
 *  <b>`x`</b>: A tensor.
-*  <b>`keep_prob`</b>: A Python float. The probability that each element is kept.
+*  <b>`keep_prob`</b>: A scalar `Tensor` with the same type as x. The probability
+    that each element is kept.
 *  <b>`noise_shape`</b>: A 1-D `Tensor` of type `int32`, representing the
     shape for randomly generated keep/drop flags.
 *  <b>`seed`</b>: A Python integer. Used to create random seeds. See
diff --git a/tensorflow/g3doc/api_docs/python/train.md b/tensorflow/g3doc/api_docs/python/train.md
index 85739e6d3b..a1ea3c2a2f 100644
--- a/tensorflow/g3doc/api_docs/python/train.md
+++ b/tensorflow/g3doc/api_docs/python/train.md
@@ -850,9 +850,9 @@ Example: decay every 100000 steps with a base of 0.96:
 ...
 global_step = tf.Variable(0, trainable=False)
 starter_learning_rate = 0.1
-learning_rate = tf.exponential_decay(starter_learning_rate, global_step,
-                                     100000, 0.96, staircase=True)
-optimizer = tf.GradientDescent(learning_rate)
+learning_rate = tf.train.exponential_decay(starter_learning_rate, global_step,
+                                           100000, 0.96, staircase=True)
+optimizer = tf.GradientDescentOptimizer(learning_rate)
 # Passing global_step to minimize() will increment it at each step.
 optimizer.minimize(...my loss..., global_step=global_step)
 ```
@@ -888,7 +888,7 @@ moving averages for evaluations often improve results significantly.
 
 ### `class tf.train.ExponentialMovingAverage` <a class="md-anchor" id="ExponentialMovingAverage"></a>
 
-Maintains moving averages of variables by employing and exponential decay.
+Maintains moving averages of variables by employing an exponential decay.
 
 When training a model, it is often beneficial to maintain moving averages of
 the trained parameters.  Evaluations that use averaged parameters sometimes
diff --git a/tensorflow/g3doc/get_started/leftnav_files b/tensorflow/g3doc/get_started/leftnav_files
new file mode 100644
index 0000000000..3fccbc0db5
--- /dev/null
+++ b/tensorflow/g3doc/get_started/leftnav_files
@@ -0,0 +1,3 @@
+index.md
+os_setup.md
+basic_usage.md
\ No newline at end of file
diff --git a/tensorflow/g3doc/how_tos/leftnav_files b/tensorflow/g3doc/how_tos/leftnav_files
new file mode 100644
index 0000000000..f2b0a9fe9d
--- /dev/null
+++ b/tensorflow/g3doc/how_tos/leftnav_files
@@ -0,0 +1,10 @@
+variables/index.md
+../tutorials/mnist/tf/index.md
+summaries_and_tensorboard/index.md
+graph_viz/index.md
+reading_data/index.md
+threading_and_queues/index.md
+adding_an_op/index.md
+new_data_formats/index.md
+using_gpu/index.md
+variable_scope/index.md
\ No newline at end of file
diff --git a/tensorflow/g3doc/resources/leftnav_files b/tensorflow/g3doc/resources/leftnav_files
new file mode 100644
index 0000000000..2e1940b5d4
--- /dev/null
+++ b/tensorflow/g3doc/resources/leftnav_files
@@ -0,0 +1,5 @@
+bib.md
+uses.md
+faq.md
+glossary.md
+dims_types.md
diff --git a/tensorflow/g3doc/tutorials/leftnav_files b/tensorflow/g3doc/tutorials/leftnav_files
new file mode 100644
index 0000000000..d27ec1bbf6
--- /dev/null
+++ b/tensorflow/g3doc/tutorials/leftnav_files
@@ -0,0 +1,10 @@
+mnist/beginners/index.md
+mnist/pros/index.md
+mnist/tf/index.md
+deep_cnn/index.md
+word2vec/index.md
+recurrent/index.md
+seq2seq/index.md
+mandelbrot/index.md
+pdes/index.md
+mnist/download/index.md
diff --git a/tensorflow/g3doc/tutorials/mnist/beginners/index.md b/tensorflow/g3doc/tutorials/mnist/beginners/index.md
index fcd891c58b..f21c97b7c6 100644
--- a/tensorflow/g3doc/tutorials/mnist/beginners/index.md
+++ b/tensorflow/g3doc/tutorials/mnist/beginners/index.md
@@ -418,6 +418,6 @@ a look at this
 [list of results](http://rodrigob.github.io/are_we_there_yet/build/classification_datasets_results.html).)
 
 What matters is that we learned from this model. Still, if you're feeling a bit
-down about these results, check out [the next tutorial](../../../tutorials/index.md) where we
+down about these results, check out [the next tutorial](../../../tutorials/mnist/pros/index.md) where we
 do a lot better, and learn how to build more sophisticated models using
 TensorFlow!
diff --git a/tensorflow/g3doc/tutorials/mnist/pros/index.md b/tensorflow/g3doc/tutorials/mnist/pros/index.md
index 0a7d1aeae0..58cec4c348 100644
--- a/tensorflow/g3doc/tutorials/mnist/pros/index.md
+++ b/tensorflow/g3doc/tutorials/mnist/pros/index.md
@@ -232,9 +232,9 @@ print accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels})
 ## Build a Multilayer Convolutional Network <a class="md-anchor" id="AUTOGENERATED-build-a-multilayer-convolutional-network"></a>
 
 Getting 91% accuracy on MNIST is bad. It's almost embarrassingly bad. In this
-section, we'll fix that, jumping from a very simple model to something moderately
-sophisticated: a small convolutional neural network. This will get us to around
-99.2% accuracy -- not state of the art, but respectable.
+section, we'll fix that, jumping from a very simple model to something
+moderately sophisticated: a small convolutional neural network. This will get us
+to around 99.2% accuracy -- not state of the art, but respectable.
 
 ### Weight Initialization <a class="md-anchor" id="AUTOGENERATED-weight-initialization"></a>
 
diff --git a/tensorflow/g3doc/tutorials/recurrent/index.md b/tensorflow/g3doc/tutorials/recurrent/index.md
index d1be50e00f..510ecd5d66 100644
--- a/tensorflow/g3doc/tutorials/recurrent/index.md
+++ b/tensorflow/g3doc/tutorials/recurrent/index.md
@@ -186,7 +186,7 @@ Now we can run the model:
 
 ```
 bazel-bin/tensorflow/models/rnn/ptb/ptb_word_lm \
-  --data_path=/tmp/simple-examples/data/ --alsologtostderr --model small
+  --data_path=/tmp/simple-examples/data/ --model small
 ```
 
 There are 3 supported model configurations in the tutorial code: "small",
diff --git a/tensorflow/g3doc/tutorials/seq2seq/index.md b/tensorflow/g3doc/tutorials/seq2seq/index.md
index a13f7cd95c..66c653e1f8 100644
--- a/tensorflow/g3doc/tutorials/seq2seq/index.md
+++ b/tensorflow/g3doc/tutorials/seq2seq/index.md
@@ -12,7 +12,7 @@ This tutorial will show you how to build and train such a system end-to-end.
 You can start by running this binary.
 
 ```
-bazel run -c opt <...>/models/rnn/translate/translate.py
+bazel run -c opt <...>/models/rnn/translate:translate
   --data_dir [your_data_directory]
 ```
 
diff --git a/tensorflow/models/embedding/BUILD b/tensorflow/models/embedding/BUILD
index fe52778fa9..9cd0d24b5b 100644
--- a/tensorflow/models/embedding/BUILD
+++ b/tensorflow/models/embedding/BUILD
@@ -49,6 +49,9 @@ py_test(
     size = "small",
     srcs = ["word2vec_optimized_test.py"],
     srcs_version = "PY2AND3",
+    tags = [
+        "notsan",
+    ],
     deps = [
         ":word2vec_optimized",
         "//tensorflow:tensorflow_py",
diff --git a/tensorflow/models/embedding/word2vec.py b/tensorflow/models/embedding/word2vec.py
index 29bd9602bb..400168c8ca 100644
--- a/tensorflow/models/embedding/word2vec.py
+++ b/tensorflow/models/embedding/word2vec.py
@@ -18,13 +18,14 @@ from __future__ import division
 from __future__ import print_function
 
 import os
-from six.moves import xrange  # pylint: disable=redefined-builtin
 import sys
 import threading
 import time
 
 import tensorflow.python.platform
 
+from six.moves import xrange  # pylint: disable=redefined-builtin
+
 import numpy as np
 import tensorflow as tf
 
diff --git a/tensorflow/models/embedding/word2vec_optimized.py b/tensorflow/models/embedding/word2vec_optimized.py
index e6fccf689a..547abf8c3e 100644
--- a/tensorflow/models/embedding/word2vec_optimized.py
+++ b/tensorflow/models/embedding/word2vec_optimized.py
@@ -17,13 +17,14 @@ from __future__ import division
 from __future__ import print_function
 
 import os
-from six.moves import xrange  # pylint: disable=redefined-builtin
 import sys
 import threading
 import time
 
 import tensorflow.python.platform
 
+from six.moves import xrange  # pylint: disable=redefined-builtin
+
 import numpy as np
 import tensorflow as tf
 
diff --git a/tensorflow/models/rnn/ptb/ptb_word_lm.py b/tensorflow/models/rnn/ptb/ptb_word_lm.py
index 8a6c7203c2..3a7330836b 100644
--- a/tensorflow/models/rnn/ptb/ptb_word_lm.py
+++ b/tensorflow/models/rnn/ptb/ptb_word_lm.py
@@ -37,8 +37,7 @@ To compile on GPU:
   bazel build -c opt tensorflow --config=cuda \
     tensorflow/models/rnn/ptb:ptb_word_lm
 To run:
-  ./bazel-bin/.../ptb_word_lm \
-    --data_path=/tmp/simple-examples/data/ --alsologtostderr
+  ./bazel-bin/.../ptb_word_lm --data_path=/tmp/simple-examples/data/
 
 """
 from __future__ import absolute_import
diff --git a/tensorflow/models/rnn/rnn_cell.py b/tensorflow/models/rnn/rnn_cell.py
index ff93b47892..3ffff7815c 100644
--- a/tensorflow/models/rnn/rnn_cell.py
+++ b/tensorflow/models/rnn/rnn_cell.py
@@ -68,8 +68,8 @@ class RNNCell(object):
       A 2D Tensor of shape [batch_size x state_size] filled with zeros.
     """
     zeros = tf.zeros(tf.pack([batch_size, self.state_size]), dtype=dtype)
-    # The reshape below is a no-op, but it allows shape inference of shape[1].
-    return tf.reshape(zeros, [-1, self.state_size])
+    zeros.set_shape([None, self.state_size])
+    return zeros
 
 
 class BasicRNNCell(RNNCell):
diff --git a/tensorflow/python/client/session_test.py b/tensorflow/python/client/session_test.py
index 96b7136b15..99bf356cc2 100644
--- a/tensorflow/python/client/session_test.py
+++ b/tensorflow/python/client/session_test.py
@@ -602,6 +602,29 @@ class SessionTest(test_util.TensorFlowTestCase):
       self.assertEqual(45.0, sess.run(b'f:0'))
       self.assertEqual(45.0, sess.run(r'f:0'))
 
+  def testIncorrectGraph(self):
+    with ops.Graph().as_default() as g_1:
+      c_1 = constant_op.constant(1.0, name='c')
+
+    with ops.Graph().as_default() as g_2:
+      c_2 = constant_op.constant(2.0, name='c')
+
+    self.assertEqual('c', c_1.op.name)
+    self.assertEqual('c', c_2.op.name)
+
+    with session.Session(graph=g_1) as sess_1:
+      self.assertEqual(1.0, sess_1.run(c_1))
+      with self.assertRaises(ValueError):
+        sess_1.run(c_2)
+      with self.assertRaises(ValueError):
+        sess_1.run(c_2.op)
+
+    with session.Session(graph=g_2) as sess_2:
+      with self.assertRaises(ValueError):
+        sess_2.run(c_1)
+      with self.assertRaises(ValueError):
+        sess_2.run(c_1.op)
+      self.assertEqual(2.0, sess_2.run(c_2))
 
 if __name__ == '__main__':
   googletest.main()
diff --git a/tensorflow/python/framework/ops.py b/tensorflow/python/framework/ops.py
index 54917a506f..e5a8f8a4dc 100644
--- a/tensorflow/python/framework/ops.py
+++ b/tensorflow/python/framework/ops.py
@@ -1552,7 +1552,7 @@ class Graph(object):
     # new operations can be added.
     self._finalized = False
     # Functions defined in the graph
-    self._functions = []
+    self._functions = collections.OrderedDict()
 
   def _check_not_finalized(self):
     """Check if the graph is finalized.
@@ -1647,6 +1647,9 @@ class Graph(object):
     Returns:
       A [`GraphDef`](https://tensorflow.googlesource.com/tensorflow/+/master/tensorflow/core/framework/graph.proto)
       protocol buffer.
+
+    Raises:
+      ValueError: If the graph_def would be too large.
     """
     graph = graph_pb2.GraphDef()
     bytesize = 0
@@ -1658,18 +1661,18 @@ class Graph(object):
         if bytesize >= (1 << 31) or bytesize < 0:
           raise ValueError("GraphDef cannot be larger than 2GB.")
     if self._functions:
-      for f in self._functions:
+      for f in self._functions.values():
         bytesize += f.ByteSize()
         if bytesize >= (1 << 31) or bytesize < 0:
           raise ValueError("GraphDef cannot be larger than 2GB.")
-      graph.library.function.extend(self._functions)
+      graph.library.function.extend(self._functions.values())
     return graph
 
   def _add_function(self, function_def):
     """Adds a function to the graph.
 
     The function is specified as a [`FunctionDef`]
-    (https://tensorflow.googlesource.com/tensorflow/+/master/tensorflow/core/framework/graph.proto)
+    (https://tensorflow.googlesource.com/tensorflow/+/master/tensorflow/core/framework/function.proto)
     protocol buffer.
 
     After the function has been added, you can call to the function by
@@ -1679,7 +1682,14 @@ class Graph(object):
     Args:
       function_def: A `FunctionDef` protocol buffer.
     """
-    self._functions.append(function_def)
+    previous_def = self._functions.get(function_def.signature.name, None)
+    if previous_def:
+      if previous_def != function_def:
+        raise ValueError("Another function is already defined with that name")
+      else:
+        # No need to add again.
+        return
+    self._functions[function_def.signature.name] = function_def
 
   # Helper functions to create operations.
   def create_op(self, op_type, inputs, dtypes,
@@ -1871,9 +1881,13 @@ class Graph(object):
 
     elif isinstance(obj, Tensor) and allow_tensor:
       # Actually obj is just the object it's referring to.
+      if obj.graph is not self:
+        raise ValueError("Tensor %s is not an element of this graph." % obj)
       return obj
     elif isinstance(obj, Operation) and allow_operation:
       # Actually obj is just the object it's referring to.
+      if obj.graph is not self:
+        raise ValueError("Operation %s is not an element of this graph." % obj)
       return obj
     else:
       # We give up!
diff --git a/tensorflow/python/kernel_tests/gradient_checker.py b/tensorflow/python/kernel_tests/gradient_checker.py
index 1adf22d512..d08fc393fa 100644
--- a/tensorflow/python/kernel_tests/gradient_checker.py
+++ b/tensorflow/python/kernel_tests/gradient_checker.py
@@ -80,7 +80,7 @@ def _ComputeTheoricalJacobian(x, x_shape, x_data, dy, dy_shape, dx):
 def _ComputeNumericJacobian(x, x_shape, x_data, y, y_shape, delta):
   """Computes the numeric Jacobian for dy/dx.
 
-  Computes the numeric Japcobian by slightly perturbing the inputs and
+  Computes the numeric Jacobian by slightly perturbing the inputs and
   measuring the differences on the output.
 
   Args:
diff --git a/tensorflow/python/ops/array_ops.py b/tensorflow/python/ops/array_ops.py
index 78262a55f3..33fdc017c7 100644
--- a/tensorflow/python/ops/array_ops.py
+++ b/tensorflow/python/ops/array_ops.py
@@ -624,7 +624,7 @@ def placeholder(dtype, shape=None, name=None):
   For example:
 
   ```python
-  x = tf.placeholder(float, shape=(1024, 1024))
+  x = tf.placeholder(tf.float32, shape=(1024, 1024))
   y = tf.matmul(x, x)
 
   with tf.Session() as sess:
diff --git a/tensorflow/python/training/input.py b/tensorflow/python/training/input.py
index 77c496fb85..3f9371fead 100644
--- a/tensorflow/python/training/input.py
+++ b/tensorflow/python/training/input.py
@@ -243,7 +243,7 @@ def _merge_shapes(shape_list, enqueue_many):
   shape_list = [tensor_shape.as_shape(s) for s in shape_list]
   if enqueue_many:
     # We want the shapes without the leading batch dimension.
-    shape_list = [s.WithRankAtLeast(1)[1:] for s in shape_list]
+    shape_list = [s.with_rank_at_least(1)[1:] for s in shape_list]
   merged_shape = shape_list[0]
   for s in shape_list[1:]:
     merged_shape.merge_with(s)
diff --git a/tensorflow/python/training/input_test.py b/tensorflow/python/training/input_test.py
index d85df52431..bd6e199054 100644
--- a/tensorflow/python/training/input_test.py
+++ b/tensorflow/python/training/input_test.py
@@ -269,6 +269,31 @@ class BatchTest(tf.test.TestCase):
       for thread in threads:
         thread.join()
 
+  def testOneThreadEnqueueMany(self):
+    with self.test_session() as sess:
+      batch_size = 10
+      num_batches = 3
+      zero64 = tf.constant(0, dtype=tf.int64)
+      examples = tf.Variable(zero64)
+      counter = examples.count_up_to(num_batches * batch_size)
+      pre_batched = tf.train.batch([counter, "string"], batch_size=2)
+      batched = tf.train.batch(pre_batched, enqueue_many=True,
+                               batch_size=batch_size)
+      tf.initialize_all_variables().run()
+      threads = tf.train.start_queue_runners()
+
+      for i in range(num_batches):
+        results = sess.run(batched)
+        self.assertAllEqual(results[0], np.arange(i * batch_size,
+                                                  (i + 1) * batch_size))
+        self.assertAllEqual(results[1], ["string"] * batch_size)
+
+      # Reached the limit.
+      with self.assertRaises(tf.errors.OutOfRangeError):
+        sess.run(batched)
+      for thread in threads:
+        thread.join()
+
   def testManyThreads(self):
     with self.test_session() as sess:
       batch_size = 10
diff --git a/tensorflow/python/training/moving_averages.py b/tensorflow/python/training/moving_averages.py
index d834b698ba..50a8f9bbce 100644
--- a/tensorflow/python/training/moving_averages.py
+++ b/tensorflow/python/training/moving_averages.py
@@ -45,7 +45,7 @@ def assign_moving_average(variable, value, decay, name=None):
 
 
 class ExponentialMovingAverage(object):
-  """Maintains moving averages of variables by employing and exponential decay.
+  """Maintains moving averages of variables by employing an exponential decay.
 
   When training a model, it is often beneficial to maintain moving averages of
   the trained parameters.  Evaluations that use averaged parameters sometimes
diff --git a/tensorflow/tensorboard/components/tf-graph-common/lib/hierarchy.ts b/tensorflow/tensorboard/components/tf-graph-common/lib/hierarchy.ts
index 8ada82770c..36692dd0f0 100644
--- a/tensorflow/tensorboard/components/tf-graph-common/lib/hierarchy.ts
+++ b/tensorflow/tensorboard/components/tf-graph-common/lib/hierarchy.ts
@@ -372,7 +372,7 @@ function findEdgeTargetsInGraph(
 
 interface HierarchyParams {
   verifyTemplate: boolean;
-  groupSeries: boolean;
+  seriesNodeMinSize: number;
 }
 
 /**
@@ -396,8 +396,8 @@ export function build(graph: tf.graph.SlimGraph, params: HierarchyParams,
   }, tracker)
   .then(() => {
     return runAsyncTask("Detect series", 20, () => {
-      if (params.groupSeries) {
-        groupSeries(h.root, h, seriesNames);
+      if (params.seriesNodeMinSize > 0) {
+        groupSeries(h.root, h, seriesNames, params.seriesNodeMinSize);
       }
     }, tracker);
   })
@@ -550,15 +550,17 @@ function addEdges(h: Hierarchy, graph: SlimGraph,
  *
  * @param metanode
  * @param hierarchy
+ * @param threshold If the series has this many nodes or more, then group them
+ *     into a series.
  * @return A dictionary from node name to series node name that contains the node
  */
 function groupSeries(metanode: Metanode, hierarchy: Hierarchy,
-    seriesNames: { [name: string]: string }) {
+    seriesNames: { [name: string]: string }, threshold: number) {
   let metagraph = metanode.metagraph;
   _.each(metagraph.nodes(), n => {
     let child = metagraph.node(n);
     if (child.type === tf.graph.NodeType.META) {
-      groupSeries(<Metanode>child, hierarchy, seriesNames);
+      groupSeries(<Metanode>child, hierarchy, seriesNames, threshold);
     }
   });
 
@@ -569,6 +571,9 @@ function groupSeries(metanode: Metanode, hierarchy: Hierarchy,
   // metagraph.
   _.each(seriesDict, function(seriesNode: SeriesNode, seriesName: string) {
     let nodeMemberNames = seriesNode.metagraph.nodes();
+    if (nodeMemberNames.length < threshold) {
+      return;
+    }
     let firstMember = seriesNode.metagraph.node(nodeMemberNames[0]);
     let seriesType = firstMember.type;
 
diff --git a/tensorflow/tensorboard/components/tf-graph-loader/tf-graph-loader.html b/tensorflow/tensorboard/components/tf-graph-loader/tf-graph-loader.html
index 3f290f2152..c3647a41b5 100644
--- a/tensorflow/tensorboard/components/tf-graph-loader/tf-graph-loader.html
+++ b/tensorflow/tensorboard/components/tf-graph-loader/tf-graph-loader.html
@@ -126,7 +126,9 @@ Polymer({
       }
       var hierarchyParams = {
         verifyTemplate: true,
-        groupSeries: true,
+        // If a set of numbered op nodes has at least this number of nodes
+        // then group them into a series node.
+        seriesNodeMinSize: 5,
       };
       var hierarchyTracker = tf.getSubtaskTracker(tracker, 50,
           'Namespace hierarchy');
diff --git a/third_party/eigen3/unsupported/Eigen/CXX11/src/NeuralNetworks/Pooling.h b/third_party/eigen3/unsupported/Eigen/CXX11/src/NeuralNetworks/Pooling.h
index 8dea22806c..942b060ba7 100644
--- a/third_party/eigen3/unsupported/Eigen/CXX11/src/NeuralNetworks/Pooling.h
+++ b/third_party/eigen3/unsupported/Eigen/CXX11/src/NeuralNetworks/Pooling.h
@@ -196,7 +196,7 @@ namespace internal {
 
 template <typename T> struct AvgPoolMeanReducer
 {
-#if (EIGEN_ARCH_i386 || EIGEN_ARCH_x86_64 || defined (EIGEN_USE_GPU) || defined(__CUDACC__) || defined(__CUDA_ARCH__))
+#if (EIGEN_ARCH_i386 || EIGEN_ARCH_x86_64) && !defined(__CUDACC__)
   // We only support packet access for floats.
   static const bool PacketAccess = internal::is_same<T, float>::value;
 #else
@@ -216,8 +216,16 @@ template <typename T> struct AvgPoolMeanReducer
     }
   }
 
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE T initialize() const {
+    return static_cast<T>(0);
+  }
+
+  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE T finalize(const T accum) const {
+    eigen_assert(scalarCount_ > 0);
+    return accum / scalarCount_;
+  }
 
-#if (!defined (EIGEN_USE_GPU) || !defined(__CUDACC__) || !defined(__CUDA_ARCH__))
+#if (EIGEN_ARCH_i386 || EIGEN_ARCH_x86_64) && !defined(__CUDACC__)
 #ifdef EIGEN_VECTORIZE_AVX
 #define pequal(a,b) _mm256_cmp_ps(a,b,_CMP_EQ_UQ)
 #define psel(a,b,false_mask) _mm256_blendv_ps(a,b,false_mask)
@@ -238,29 +246,11 @@ template <typename T> struct AvgPoolMeanReducer
     packetCount_ = padd<Packet>(packetCount_, psel(pset1<Packet>(1), pset1<Packet>(0), skip_mask));
   }
 
-#else
-#define pequal(a,b) make_float4(a.x == b.x ? 1.f : 0, a.y == b.y ? 1.f : 0, a.z == b.z ? 1.f : 0, a.w == b.w ? 1.f : 0)
-#define psel(a,b,c) make_float4(c.x ? b.x : a.x, c.y ? b.y : a.y, c.z ? b.z : a.z, c.w ? b.w : a.w)
-
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void reducePacket(const float4& p, float4* accum) {
-    float4 skip_mask = pequal(p, pset1<float4>(-Eigen::NumTraits<float>::highest()));
-    (*accum) = padd<float4>(*accum, psel(p, pset1<float4>(0), skip_mask));
-    packetCount_ = padd<float4>(packetCount_, psel(pset1<float4>(1), pset1<float4>(0), skip_mask));
-  }
-
-#endif
-
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE T initialize() const {
-    return static_cast<T>(0);
-  }
   template <typename Packet>
   EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Packet initializePacket() const {
     return pset1<Packet>(0);
   }
-  EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE T finalize(const T accum) const {
-    eigen_assert(scalarCount_ > 0);
-    return accum / scalarCount_;
-  }
+
   template <typename Packet>
   EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Packet finalizePacket(const Packet& vaccum) const {
     return pdiv(vaccum, packetCount_);
@@ -269,6 +259,7 @@ template <typename T> struct AvgPoolMeanReducer
   EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE T finalizeBoth(const T saccum, const Packet& vaccum) const {
     return (saccum + predux(vaccum)) / (scalarCount_ + predux(packetCount_));
   }
+#endif
 
  protected:
     typedef typename packet_traits<T>::type Packet;