CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  TF_DeleteStatus(status);
}
BENCHMARK(BM_ReadVariable);

TEST(CAPI, StringAttributes) {
  // Test that TFE_OpSetAttrString doesn't hold on to the value after it
  // returns.
  TF_Status* status = TF_NewStatus();
  TFE_ContextOptions* opts = TFE_NewContextOptions();
  TFE_Context* ctx = TFE_NewContext(opts, status);

  }
  ~ReluGradientFunction() override {
    for (auto output : forward_outputs_) {
      if (output) {
        output->Unref();
      }
    }
  }

 private:
  // TODO(b/174778737): Only hold needed outputs.
  vector<AbstractTensorHandle*> forward_outputs_;
};

Status BroadcastMul(AbstractContext* ctx, AbstractTensorHandle* vec,
                    AbstractTensorHandle* mat,

  for (int i = 0; i < noutputs; ++i) {
    c_outputs[i] = nullptr;
  }
}

// TF_TensorToTensorV1 decodes a string serialization to DT_RESOURCE.
// In the TFv1 convention, TF_Tensor can hold a string serialization of
// DT_RESOURCE. The string serialization is converted back to a
// ResourceHandle during Session run where the TF_Tensor is converted to a
// Tensor.

      License. However, in accepting such obligations, You may act only
      on Your own behalf and on Your sole responsibility, not on behalf
      of any other Contributor, and only if You agree to indemnify,
      defend, and hold each Contributor harmless for any liability
      incurred by, or claims asserted against, such Contributor by reason

  }
  ~MatMulGradientFunction() override {
    for (auto input : forward_inputs_) {
      if (input) {
        input->Unref();
      }
    }
  }

 private:
  // TODO(b/174778737): Only hold needed inputs.
  vector<AbstractTensorHandle*> forward_inputs_;
  AttrBuilder forward_attrs_;
};

class NegGradientFunction : public GradientFunction {
 public:
  Status Compute(AbstractContext* ctx,

#include <memory>

#include "tensorflow/c/experimental/filesystem/filesystem_interface.h"
#include "tensorflow/core/platform/file_system.h"

/// This file builds classes needed to hold a filesystem implementation in the
/// modular world. Once all TensorFlow filesystems are converted to use the
/// plugin based approach, this file will replace the one in core/platform and

 public:
  using GradientTape<AbstractTensorHandle, GradientFunction,
                     TapeTensor>::GradientTape;
  // Returns whether the tape is persistent, i.e., whether the tape will hold
  // onto its internal state after a call to `ComputeGradient`.
  using GradientTape<AbstractTensorHandle, GradientFunction,
                     TapeTensor>::IsPersistent;

    return ptr->getKind() == kGraph;
  }
  ~GraphOperation() override {}

 private:
  friend class GraphContext;  // For access to op_.
  TF_Graph* g_;
  std::unique_ptr<TF_OperationDescription> op_;
  // Hold `op_type` and `op_name` till both are available since we need both
  // to build a graph operation.
  string op_type_;
  const char* op_name_ = nullptr;
  // TODO(srbs): Use this.
  string device_name_;
};

///   3. Versioning metadata;
///   4. Plugin registration API and the DSO entry point.

#ifdef __cplusplus
extern "C" {
#endif  // __cplusplus

/// SECTION 1. Opaque data structures to hold plugin specific data
/// ----------------------------------------------------------------------------
///
/// The following data structures incorporate a `void*` that is opaque to

*   `tf.keras.optimizers.Optimizer` now points to the new Keras optimizer, and
    old optimizers have moved to the `tf.keras.optimizers.legacy` namespace.
    If you find your workflow failing due to this change,
    you may be facing one of the following issues:

    *   **Checkpoint loading failure.** The new optimizer handles optimizer
        state differently from the old optimizer, which simplies the logic of