[tf.lite] Remove Interpreter::UseNNAPI

This method was deprecated and is no longer supported.  Clients should
use the more stable method of applyiing the delegate manually:

  `Interpreter::ModifyGraphWithDelegate(NnApiDelegate())`

PiperOrigin-RevId: 339319268
Change-Id: I687baa9e2b380eac58c1e882cd1692a978375ec2

Author: jdduke

RELEASE.md

@@ -28,7 +28,13 @@
     *   Exposing `tf.data.experimental.ExternalStatePolicy`, which can be used
         to control how external state should be handled during dataset
         serialization or iterator checkpointing.
-  
+
+*   `tf.lite`:
+    *   NNAPI
+        *   Removed deprecated `Interpreter::UseNNAPI(bool)` C++ API.
+            *   Use `NnApiDelegate()` and related delegate configuration methods
+                directly.
+
 ## Thanks to our Contributors
 
 This release contains contributions from many people at Google, as well as:

tensorflow/lite/BUILD

@@ -261,10 +261,8 @@ cc_library(
         "//tensorflow/lite/core/api",
         "//tensorflow/lite/core/api:verifier",
         "//tensorflow/lite/delegates:status",
-        "//tensorflow/lite/delegates/nnapi:nnapi_delegate",
         "//tensorflow/lite/experimental/resource",
         "//tensorflow/lite/kernels/internal:compatibility",
-        "//tensorflow/lite/nnapi:nnapi_implementation",
         "//tensorflow/lite/profiling:platform_profiler",
         "//tensorflow/lite/schema:schema_fbs",
         "//tensorflow/lite/schema:schema_utils",
@@ -296,9 +294,7 @@ cc_library(
         "//tensorflow/lite/c:common",
         "//tensorflow/lite/core/api",
         "//tensorflow/lite/core/api:verifier",
-        "//tensorflow/lite/delegates/nnapi:nnapi_delegate",
         "//tensorflow/lite/experimental/resource",
-        "//tensorflow/lite/nnapi:nnapi_implementation",
         "//tensorflow/lite/schema:schema_fbs",
     ],
 )

tensorflow/lite/core/subgraph.cc

@@ -23,7 +23,6 @@ limitations under the License.
 #include "tensorflow/lite/c/common.h"
 #include "tensorflow/lite/context_util.h"
 #include "tensorflow/lite/core/api/tensor_utils.h"
-#include "tensorflow/lite/delegates/nnapi/nnapi_delegate.h"
 #include "tensorflow/lite/graph_info.h"
 #include "tensorflow/lite/minimal_logging.h"
 #include "tensorflow/lite/schema/schema_generated.h"
@@ -991,13 +990,6 @@ TfLiteStatus Subgraph::Invoke() {
     return kTfLiteError;
   }
 
-  // This is only needed for UseNNAPI(true);
-  if (should_apply_nnapi_delegate_ && !applied_nnapi_delegate_) {
-    TF_LITE_ENSURE_OK(&context_, ModifyGraphWithDelegate(NnApiDelegate()));
-    // only need to modify the graph once upon the first invocation.
-    applied_nnapi_delegate_ = true;
-  }
-
   // Invocations are always done in node order.
   // Note that calling Invoke repeatedly will cause the original memory plan to
   // be reused, unless either ResizeInputTensor() or AllocateTensors() has been
@@ -1335,16 +1327,6 @@ TfLiteStatus Subgraph::ResizeTensorImpl(TfLiteTensor* tensor,
   return kTfLiteOk;
 }
 
-void Subgraph::UseNNAPI(bool enable) {
-  // Note that there is no way to disable the delegate once it modified the
-  // graph.
-  if (applied_nnapi_delegate_ && !enable) {
-    ReportError("Attempting to disable NNAPI delegate after it's applied.");
-  } else {
-    should_apply_nnapi_delegate_ = enable;
-  }
-}
-
 void Subgraph::SwitchToDelegateContext() {
   context_.GetNodeAndRegistration = GetNodeAndRegistration;
   context_.ReplaceNodeSubsetsWithDelegateKernels =

tensorflow/lite/core/subgraph.h

@@ -25,16 +25,12 @@ limitations under the License.
 #include "tensorflow/lite/c/common.h"
 #include "tensorflow/lite/core/api/profiler.h"
 #include "tensorflow/lite/core/macros.h"
-#include "tensorflow/lite/delegates/nnapi/nnapi_delegate.h"
 #include "tensorflow/lite/experimental/resource/resource_base.h"
 #include "tensorflow/lite/memory_planner.h"
 #include "tensorflow/lite/util.h"
 
 namespace tflite {
 
-// Forward declare since NNAPIDelegate uses Interpreter.
-class NNAPIDelegate;
-
 class Subgraph {
  public:
   friend class Interpreter;
@@ -247,8 +243,6 @@ class Subgraph {
   // Entry point for C node plugin API to report an error.
   void ReportError(const char* format, ...);
 
-  void UseNNAPI(bool enable);
-
   // Return the subgraph specific context.
   TfLiteContext* context() { return &context_; }
 
@@ -704,10 +698,6 @@ class Subgraph {
   // Used by PreviewDelegateParitioning.
   std::vector<TfLiteDelegateParams> partitioning_preview_cache_;
 
-  // Whether to use delegate to modify the graph.
-  bool should_apply_nnapi_delegate_ = false;
-  bool applied_nnapi_delegate_ = false;
-
   std::unique_ptr<MemoryPlanner> memory_planner_;
 
   // Contains <tensor idx, custom allocation> pairs for all applicable tensors.

tensorflow/lite/g3doc/guide/faq.md

@@ -135,11 +135,9 @@ like this:
     to do this. However, increasing threads results in performance variability
     depending on the environment.
 *   *Use Hardware Accelerators.* TensorFlow Lite supports model acceleration for
-    specific hardware using delegates. For example, to use Android’s Neural
-    Networks API, call
-    [`UseNNAPI`](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/interpreter.h#L343)
-    on the interpreter. Or take a look at our
-    [GPU delegate tutorial](../performance/gpu.md).
+    specific hardware using delegates. See our
+    [Delegates](../performance/delegates.md) guide for information on what
+    accelerators are supported and how to use them with your model on-device.
 *   *(Advanced) Profile Model.* The Tensorflow Lite
     [benchmarking tool](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/tools/benchmark)
     has a built-in profiler that can show per-operator statistics. If you know

tensorflow/lite/g3doc/performance/best_practices.md

@@ -107,9 +107,8 @@ interpreter execution. TensorFlow Lite can use delegates by:
 *   Using Android's
     [Neural Networks API](https://developer.android.com/ndk/guides/neuralnetworks/).
     You can utilize these hardware accelerator backends to improve the speed and
-    efficiency of your model. To enable the Neural Networks API, call
-    [UseNNAPI](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/interpreter.h#L343)
-    on the interpreter instance.
+    efficiency of your model. To enable the Neural Networks API, check out
+    the [NNAPI delegate](nnapi.md) guide.
 *   GPU delegate is available on Android and iOS, using OpenGL/OpenCL and Metal,
     respectively. To try them out, see the [GPU delegate tutorial](gpu.md) and
     [documentation](gpu_advanced.md).

tensorflow/lite/interpreter.cc

@@ -109,8 +109,6 @@ Interpreter::Interpreter(ErrorReporter* error_reporter)
   own_external_cpu_backend_context_.reset(new ExternalCpuBackendContext());
   external_contexts_[kTfLiteCpuBackendContext] =
       own_external_cpu_backend_context_.get();
-
-  primary_subgraph().UseNNAPI(false);
 }
 
 Interpreter::~Interpreter() {
@@ -344,13 +342,6 @@ TfLiteStatus Interpreter::SetExecutionPlan(const std::vector<int>& new_plan) {
   return primary_subgraph().SetExecutionPlan(new_plan);
 }
 
-void Interpreter::UseNNAPI(bool enable) {
-  TFLITE_LOG_PROD_ONCE(TFLITE_LOG_INFO,
-                       "Interpreter::UseNNAPI() is deprecated. Use "
-                       "tflite::NnApiDelegate() directly instead.");
-  primary_subgraph().UseNNAPI(enable);
-}
-
 TfLiteStatus Interpreter::SetNumThreads(int num_threads) {
   if (num_threads < -1) {
     context_->ReportError(context_,

tensorflow/lite/interpreter.h

@@ -429,15 +429,6 @@ class Interpreter {
   /// Returns status of success or failure.
   TfLiteStatus Invoke();
 
-  /// Enable or disable NNAPI (true to enable). Disabled by default.
-  ///
-  /// WARNING: NNAPI cannot be disabled after the graph has been prepared
-  /// (via `AllocateTensors`) with NNAPI enabled.
-  ///
-  /// WARNING: This API is deprecated, prefer using the NNAPI delegate directly.
-  /// This method will be removed in a future release.
-  void UseNNAPI(bool enable);
-
   /// Set the number of threads available to the interpreter.
   ///
   /// NOTE: num_threads should be >= -1.

tensorflow/lite/interpreter_test.cc

@@ -985,17 +985,6 @@ TEST(BasicInterpreter, TestOverflow) {
   }
 }
 
-TEST(BasicInterpreter, TestUseNNAPI) {
-  TestErrorReporter reporter;
-  Interpreter interpreter(&reporter);
-  interpreter.UseNNAPI(true);
-  ASSERT_EQ(interpreter.AllocateTensors(), kTfLiteOk);
-  ASSERT_EQ(interpreter.Invoke(), kTfLiteOk);
-  interpreter.UseNNAPI(false);
-  ASSERT_EQ(reporter.error_messages(),
-            "Attempting to disable NNAPI delegate after it's applied.");
-}
-
 TEST(BasicInterpreter, TestUnsupportedDelegateFunctions) {
   Interpreter interpreter;
   ASSERT_EQ(interpreter.AddTensors(2), kTfLiteOk);