Merge changes from github.

END_PUBLIC

---
Commit 1e1b3d902 authored by Pete Warden<pete@petewarden.com>
Committed by gunan<gunan@google.com>:
Changed output directory for Pi CI build to fix permissions problem with nightlies (#13257)

* Fix for RTLD_GLOBAL breakage of Pi builds, and removed Eigen version change for Pi that's no longer needed

* Fixed Pi Zero OpenBLAS build problems and tidied up directories used

* More robust checks in Pi build script

* Changed output directory for Pi CI build to fix permissions problem

---
Commit fe3a2e65c authored by Yan Facai (???)<facai.yan@gmail.com>
Committed by drpngx<drpngx@users.noreply.github.com>:
check invalid string type for dest_nodes in extract_sub_graph (#13057)

* BUG: check str type

* TST: add unit test

* CLN: remove list check

* CLN: use warning

* CLN: 2 indent

* CLN: raise TypeError if not list

* CLN: check string only

---
Commit 225ab7629 authored by Jean Wanka<jm.wanka@gmail.com>
Committed by Jean Wanka<jm.wanka@gmail.com>:
Fix polynomial decay with cycle for global step=0

For polynomial decay with cycle=True the learning rate at
step 0 becomes NaN, because in the process of calculating it we
devide by 0. This change should fix it, by setting the multiplier
for the decay steps to one for global_step=0.

---
Commit 286f57061 authored by Bjarke Hammersholt Roune<broune@google.com>
Committed by TensorFlower Gardener<gardener@tensorflow.org>:
Make Service::TransferToClient not attempt to manipulate the literal when the transfer failed, preventing a crash and allowing the caller to see the reason for the failed transfer.

PiperOrigin-RevId: 169770126

---
Commit e0501bc4d authored by Yong Tang<yong.tang.github@outlook.com>
Committed by Shanqing Cai<cais@google.com>:
Fix GRUBlockCell parameter naming inconsistency (#13153)

* Fix GRUBlockCell parameter naming inconsistency

This fix tries to fix the issue in 13137 where
parameter `cell_size` is used instead of `num_units`.
This is inconsistent with other RNN cells.

This fix adds support of `num_units` while at the same
time maintains backward compatiblility for `cell_size`.

This fix fixes 13137.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>

* Add `@deprecated_args` for 'cell_size' in `GRUBlockCell`

This commit adds `@deprecated_args` for 'cell_size' in `GRUBlockCell`

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>

* Address review comment

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>

---
Commit 02a2eba05 authored by Pete Warden<pete@petewarden.com>
Committed by gunan<gunan@google.com>:
Fix for RTLD_GLOBAL breakage of Pi builds, and removed Eigen version change that's no longer needed (#13251)

* Fix for RTLD_GLOBAL breakage of Pi builds, and removed Eigen version change for Pi that's no longer needed

* Fixed Pi Zero OpenBLAS build problems and tidied up directories used

* More robust checks in Pi build script

---
Commit 8ef722253 authored by Sanjoy Das<sanjoy@google.com>
Committed by TensorFlower Gardener<gardener@tensorflow.org>:
Remove a redundant setName.

The EmitComputation should have emitted a function with the right name, so use a
CHECK instead.

PiperOrigin-RevId: 169764856

---
Commit 1b94147dc authored by Neal Wu<wun@google.com>
Committed by TensorFlower Gardener<gardener@tensorflow.org>:
Fix broken GitHub links in tensorflow and tensorflow_models resulting from The Great Models Move (a.k.a. the research subfolder)

PiperOrigin-RevId: 169763373

---
Commit b1ada5f0c authored by Justine Tunney<jart@google.com>
Committed by TensorFlower Gardener<gardener@tensorflow.org>:
Fix TensorBoard python -m invoke in docs

PiperOrigin-RevId: 169758752

---
Commit 2957cd894 authored by Mustafa Ispir<ispir@google.com>
Committed by TensorFlower Gardener<gardener@tensorflow.org>:
Local run option of estimator training.

PiperOrigin-RevId: 169756384

---
Commit 1dc2fe7ac authored by Gunhan Gulsoy<gunan@google.com>
Committed by TensorFlower Gardener<gardener@tensorflow.org>:
BEGIN_PUBLIC
Automated g4 rollback of changelist 166264198

PiperOrigin-RevId: 169998124

1 files changed, 41 insertions, 24 deletions

diff --git a/tensorflow/stream_executor/dnn.h b/tensorflow/stream_executor/dnn.h
index 0a4525c1b7..b11c6417be 100644
--- a/tensorflow/stream_executor/dnn.h
+++ b/tensorflow/stream_executor/dnn.h
@@ -667,9 +667,26 @@ class PoolingDescriptor {
   std::vector<int64> strides_;
 };
 
-typedef int64 AlgorithmType;
-constexpr AlgorithmType kDefaultAlgorithm = -1;
-constexpr AlgorithmType kNoSuitableAlgorithmFound = -2;
+// Collects parameters for DNN algorithms
+class AlgorithmDesc {
+ public:
+  typedef int64 Index;
+  AlgorithmDesc() : algo_(kDefaultAlgorithm), tensor_ops_enabled_(false) {}
+  AlgorithmDesc(Index a, bool use_tensor_ops)
+      : algo_(a), tensor_ops_enabled_(use_tensor_ops) {}
+  bool is_default() const { return algo_ == kDefaultAlgorithm; }
+  bool tensor_ops_enabled() const { return tensor_ops_enabled_; }
+  Index algo_id() const { return algo_; }
+  bool operator==(const AlgorithmDesc& other) const {
+    return this->algo_ == other.algo_ &&
+           this->tensor_ops_enabled_ == other.tensor_ops_enabled_;
+  }
+
+ private:
+  enum { kDefaultAlgorithm = -1 };
+  Index algo_;
+  bool tensor_ops_enabled_;
+};
 
 // Describes the result from a perf experiment.
 //
@@ -679,16 +696,16 @@ constexpr AlgorithmType kNoSuitableAlgorithmFound = -2;
 class ProfileResult {
  public:
   bool is_valid() const {
-    return (algorithm_ != kDefaultAlgorithm &&
+    return (!algorithm_.is_default() &&
             elapsed_time_in_ms_ != std::numeric_limits<float>::max());
   }
-  AlgorithmType algorithm() const { return algorithm_; }
-  void set_algorithm(AlgorithmType val) { algorithm_ = val; }
+  AlgorithmDesc algorithm() const { return algorithm_; }
+  void set_algorithm(AlgorithmDesc val) { algorithm_ = val; }
   float elapsed_time_in_ms() const { return elapsed_time_in_ms_; }
   void set_elapsed_time_in_ms(float val) { elapsed_time_in_ms_ = val; }
 
  private:
-  AlgorithmType algorithm_ = kDefaultAlgorithm;
+  AlgorithmDesc algorithm_;
   float elapsed_time_in_ms_ = std::numeric_limits<float>::max();
 };
 
@@ -700,17 +717,14 @@ class ProfileResult {
 //    the allocation for the scratch memory fails.
 class AlgorithmConfig {
  public:
-  AlgorithmConfig()
-      : algorithm_(kDefaultAlgorithm),
-        algorithm_no_scratch_(kDefaultAlgorithm) {}
-  explicit AlgorithmConfig(AlgorithmType algorithm)
-      : algorithm_(algorithm), algorithm_no_scratch_(kDefaultAlgorithm) {}
-  AlgorithmConfig(AlgorithmType algorithm, AlgorithmType algorithm_no_scratch)
+  AlgorithmConfig() {}
+  explicit AlgorithmConfig(AlgorithmDesc algorithm) : algorithm_(algorithm) {}
+  AlgorithmConfig(AlgorithmDesc algorithm, AlgorithmDesc algorithm_no_scratch)
       : algorithm_(algorithm), algorithm_no_scratch_(algorithm_no_scratch) {}
-  AlgorithmType algorithm() const { return algorithm_; }
-  void set_algorithm(AlgorithmType val) { algorithm_ = val; }
-  AlgorithmType algorithm_no_scratch() const { return algorithm_no_scratch_; }
-  void set_algorithm_no_scratch(AlgorithmType val) {
+  AlgorithmDesc algorithm() const { return algorithm_; }
+  void set_algorithm(AlgorithmDesc val) { algorithm_ = val; }
+  AlgorithmDesc algorithm_no_scratch() const { return algorithm_no_scratch_; }
+  void set_algorithm_no_scratch(AlgorithmDesc val) {
     algorithm_no_scratch_ = val;
   }
   bool operator==(const AlgorithmConfig& other) const {
@@ -723,8 +737,8 @@ class AlgorithmConfig {
   string ToString() const;
 
  private:
-  AlgorithmType algorithm_;
-  AlgorithmType algorithm_no_scratch_;
+  AlgorithmDesc algorithm_;
+  AlgorithmDesc algorithm_no_scratch_;
 };
 
 // Describes a local response normalization (LRN). LRN is used e.g. in
@@ -944,8 +958,8 @@ class DnnSupport {
   //    convolution result.
   //  scratch_allocator: un-owned, may-be-null object that may allocate scratch
   //    space in order to speed up the convolution operation.
-  //  algorithm: an integer to specify which algorithm should be used for the
-  //    operation. kDefaultAlgorithm means the system will pick an algorithm
+  //  algorithm: specifies which algorithm should be used for the
+  //    operation. If algorithm.is_default(), the system will pick an algorithm
   //    by default. The coding of the algorithm is be interpretted by the
   //    underlying implementation.
   //  output_profile_result: the output profile result for this call. The
@@ -1112,7 +1126,8 @@ class DnnSupport {
 
   // Return a list of algorithms supported by the forward convolution pass.
   virtual bool GetConvolveAlgorithms(
-      bool with_winograd_nonfused, std::vector<AlgorithmType>* out_algorithms);
+      bool with_winograd_nonfused,
+      std::vector<AlgorithmDesc::Index>* out_algorithms);
 
   // Version of DoConvolve that uses pre-quantized 8 bit coefficients.
   // coefficient_scales specifies the scaling of each column of coefficients:
@@ -1191,7 +1206,8 @@ class DnnSupport {
   // Return a list of algorithms supported by the backward convolution pass for
   // data.
   virtual bool GetConvolveBackwardDataAlgorithms(
-      bool with_winograd_nonfused, std::vector<AlgorithmType>* out_algorithms);
+      bool with_winograd_nonfused,
+      std::vector<AlgorithmDesc::Index>* out_algorithms);
 
   virtual bool DoConvolveBackwardData(
       Stream* stream, const FilterDescriptor& filter_descriptor,
@@ -1239,7 +1255,8 @@ class DnnSupport {
   // Return a list of algorithms supported by the backward convolution pass for
   // filters.
   virtual bool GetConvolveBackwardFilterAlgorithms(
-      bool with_winograd_nonfused, std::vector<AlgorithmType>* out_algorithms);
+      bool with_winograd_nonfused,
+      std::vector<AlgorithmDesc::Index>* out_algorithms);
 
   virtual bool DoConvolveBackwardFilter(
       Stream* stream, const BatchDescriptor& input_descriptor,