// Grad for Softmax Input
    TF_RETURN_IF_ERROR(BroadcastMul(
        ctx, grad_outputs[0], forward_outputs_[1],
        grad_inputs.subspan(0, 1)));  // upstream_grad * local softmax grad

    // Grad for labels is null
    grad_inputs[1] = nullptr;
    return absl::OkStatus();
  }
  ~SparseSoftmaxCrossEntropyWithLogitsGradientFunction() override {}

 private:
  vector<AbstractTensorHandle*> forward_outputs_;
};

# Some tf endpoints actually lives under Keras.
_current_module.layers = _KerasLazyLoader(
    globals(),
    submodule="__internal__.legacy.layers",
    name="layers",
    mode="v1")
if _tf_uses_legacy_keras:
  _module_dir = _module_util.get_parent_dir_for_name(
      "tf_keras.api._v1.keras.__internal__.legacy.layers")
else:
  _module_dir = _module_util.get_parent_dir_for_name(

---
name: TensorFlow Lite in Play Services issue
about: Use this template for issues with TensorFlow Lite in Google Play Services
labels: 'comp:lite-in-play-services'

---

**System information**
- Android Device information (use `adb shell getprop ro.build.fingerprint`
  if possible):
- TensorFlow Lite in Play Services SDK version (found in `build.gradle`):
- Google Play Services version

    counter = std::make_unique<
        ::tensorflow::monitoring::testing::CellReader<int64_t>>(name);
  }
  template <typename... LabelType>
  int64_t Read(const LabelType&... labels);
  std::unique_ptr<::tensorflow::monitoring::testing::CellReader<int64_t>>
      counter;
};

---
name: TensorFlow Lite Converter Issue
about: Use this template for reporting issues during model conversion to TFLite
labels: 'TFLiteConverter'

---

### 1. System information

- OS Platform and Distribution (e.g., Linux Ubuntu 16.04):
- TensorFlow installation (pip package or built from source):
- TensorFlow library (version, if pip package or github SHA, if built from source):

### 2. Code

    TFE_TensorHandle* h, TF_Status* status);

// This function will block till the operation that produces `h` has
// completed. The memory returned might alias the internal memory used by
// TensorFlow. Hence, callers should not mutate this memory (for example by
// modifying the memory region pointed to by TF_TensorData() on the returned
// TF_Tensor).
TF_CAPI_EXPORT extern TF_Tensor* TFE_TensorHandleResolve(TFE_TensorHandle* h,

start by trying one of the smaller/easier issues here i.e.
[issues with the "good first issue" label](https://github.com/tensorflow/tensorflow/labels/good%20first%20issue)
and then take a look at the
[issues with the "contributions welcome" label](https://github.com/tensorflow/tensorflow/labels/stat%3Acontributions%20welcome).
These are issues that we believe are particularly well suited for outside

    TFE_MonitoringCounter1* counter);
TF_CAPI_EXPORT extern TFE_MonitoringCounterCell* TFE_MonitoringGetCellCounter1(
    TFE_MonitoringCounter1* counter, const char* label1);

// APIs for Counter with 2 labels.
typedef struct TFE_MonitoringCounter2 TFE_MonitoringCounter2;
TF_CAPI_EXPORT extern TFE_MonitoringCounter2* TFE_MonitoringNewCounter2(
    const char* name, TF_Status* status, const char* description,

}

TEST(RamFileBlockCacheTest, ParallelReads) {
  // This fetcher won't respond until either `callers` threads are calling it
  // concurrently (at which point it will respond with success to all callers),
  // or 10 seconds have elapsed (at which point it will respond with an error).
  const int callers = 4;
  BlockingCounter counter(callers);
  auto fetcher = [&counter](const string& filename, size_t offset, size_t n,

  std::unique_ptr<char[]> buffer(new char[kPosixCopyFileBufferSize]);

  off_t offset = 0;
  int bytes_transferred = 0;
  int rc = 1;
  // When `sendfile` returns 0 we stop copying and let callers handle this.
  while (offset < size && rc > 0) {
    size_t chunk = size - offset;
    if (chunk > kPosixCopyFileBufferSize) chunk = kPosixCopyFileBufferSize;

    rc = read(src_fd, buffer.get(), chunk);