absl::Span<AbstractTensorHandle* const> inputs,
                              absl::Span<AbstractTensorHandle*> outputs) {
  Tape tape(/*persistent=*/false);
  tape.Watch(inputs[0]);  // Watch x.
  AbstractTensorHandle* exp_output;
  TF_RETURN_IF_ERROR(ops::Exp(ctx, inputs[0], &exp_output, "Exp"));
  std::unique_ptr<GradientFunction> gradient_function(
      new PassThroughGradientFunction);

      bool use_batch)
      : vspace_(vspace), use_batch_(use_batch) {
    call_state_.emplace(nullptr, false);
  }

  virtual ~ForwardAccumulator() {
    for (auto accumulated : accumulated_gradients_) {
      vspace_.DeleteGradient(accumulated.second);
    }
  }

  // Tell the forward accumulator to watch tensor_id, with a Tensor tangent

             absl::Span<AbstractTensorHandle* const> inputs,
             absl::Span<AbstractTensorHandle*> outputs) -> Status {
    Tape tape(/*persistent=*/false);
    for (size_t i{}; i < inputs.size(); ++i) {
      tape.Watch(inputs[i]);
    }
    std::vector<AbstractTensorHandle*> temp_outputs(1);
    AbstractContextPtr tape_ctx(new TapeContext(ctx, &tape, grad_registry));
    TF_RETURN_IF_ERROR(

    // Compare that the graphs match.
    GraphDef expected_gdef;
    GraphDef gdef;
    EXPECT_TRUE(GetGraphDef(expected_graph_, &expected_gdef));
    EXPECT_TRUE(GetGraphDef(graph_, &gdef));
    TF_EXPECT_GRAPH_EQ(expected_gdef, gdef);

    // Compare that the output of the gradients of both graphs match.
    RunGraphsAndCompareOutputs(grad_outputs, expected_grad_outputs);
  }

#include <algorithm>
#include <cstring>
#include <limits>
#include <memory>
#include <optional>
#include <unordered_set>
#include <utility>
#include <vector>

#include "absl/strings/match.h"
// Required for IS_MOBILE_PLATFORM
#include "tensorflow/core/platform/platform.h"  // NOLINT

#if !defined(IS_MOBILE_PLATFORM) && !defined(IS_SLIM_BUILD)
#include "tensorflow/c/experimental/filesystem/modular_filesystem.h"

your changes by running the command:

```bash
tensorflow/tools/ci_build/ci_build.sh CPU tensorflow/tools/ci_build/ci_sanity.sh
```

This will catch most license, Python coding style and BUILD file issues that
may exist in your changes.

#### Running unit tests

There are two ways to run TensorFlow unit tests.

*   `tf.data`:
    *   Graduated experimental APIs:
        * [`tf.data.Dataset.ragged_batch`](https://www.tensorflow.org/api_docs/python/tf/data/Dataset/#ragged_batch), which batches elements of `tf.data.Dataset`s into `tf.RaggedTensor`s.
        * [`tf.data.Dataset.sparse_batch`](https://www.tensorflow.org/api_docs/python/tf/data/Dataset/#sparse_batch), which batches elements of `tf.data.Dataset`s into `tf.sparse.SparseTensor`s.

selects.config_setting_group(
    name = "is_cuda_enabled_and_oss",
    match_all = [
        ":is_cuda_enabled",
        ":oss",
    ],
)

# Config setting that is satisfied when building with --config=cuda for Windows
selects.config_setting_group(
    name = "is_cuda_enabled_and_windows",
    match_all = [
        ":is_cuda_enabled",
        ":windows",
    ],
)

#include "tensorflow/c/eager/c_api_experimental.h"

#include <cstdint>
#include <memory>
#include <utility>
#include <vector>

#include "absl/container/flat_hash_map.h"
#include "absl/strings/match.h"
#include "absl/time/time.h"
#include "tensorflow/c/c_api.h"
#include "tensorflow/c/eager/c_api_internal.h"
#include "tensorflow/c/eager/tfe_context_internal.h"
#include "tensorflow/c/eager/tfe_op_internal.h"

          # metadata. gcr.io helpfully includes it in the header of the response
          # as docker-content-digest: sha256:[digest]. Note we use egrep to
          # match exactly sha256:<hash> because curl may include a ^M symbol at
          # the end of the line.
          # See https://cloud.google.com/architecture/using-container-images#exploring_image_manifests_digests_and_tags