ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);
  EXPECT_EQ(2, num_dims);
  EXPECT_EQ(2, returned_dims[0]);
  EXPECT_EQ(3, returned_dims[1]);

  // Try to fetch a shape with the wrong num_dims
  TF_GraphGetTensorShape(graph, feed_out_0, returned_dims, 5, s);
  EXPECT_EQ(TF_INVALID_ARGUMENT, TF_GetCode(s)) << TF_Message(s);

  // Try to set an invalid shape (cannot change 2x3 to a 2x5).
  dims[1] = 5;

// some input tensors are not on that device.
// LINT.IfChange
// Note: Keep in sync with internal copy of enum in eager/context.h.
typedef enum TFE_ContextDevicePlacementPolicy {
  // Running operations with input tensors on the wrong device will fail.
  TFE_DEVICE_PLACEMENT_EXPLICIT = 0,
  // Copy the tensor to the right device but log a warning.
  TFE_DEVICE_PLACEMENT_WARN = 1,
  // Silently copy the tensor, which has a performance cost since the operation

        context.get(), components, device_name, status.get());
    ASSERT_TRUE(TF_GetCode(status.get()) == TF_INVALID_ARGUMENT)
        << TF_Message(status.get());
  }

  {
    // Try to extract the wrong number of components from a parallel tensor
    std::array<TFE_TensorHandle*, 1> correct_components{value_one.get()};
    TensorHandlePtr combined_value = CreatePerDeviceValues(

// case though).
//
// If the evaluation is successful, this function returns true and `output`s
// value is returned in `result`. Otherwise returns false. An error status is
// returned if something is wrong with the graph or input. Note that this may
// return false even if no error status is set.
TF_CAPI_EXPORT extern unsigned char TF_TryEvaluateConstant(TF_Graph* graph,

build:windows --host_linkopt=/DEBUG
build:windows --linkopt=/OPT:REF
build:windows --host_linkopt=/OPT:REF
build:windows --linkopt=/OPT:ICF
build:windows --host_linkopt=/OPT:ICF

# Verbose failure logs when something goes wrong
build:windows --verbose_failures

# Work around potential issues with large command lines on windows.
# See: https://github.com/bazelbuild/bazel/issues/5163
build:windows --features=compiler_param_file

- Provide links to your TensorFlow model and (optionally) TensorFlow Lite Model.
```

### 3. Failure after conversion
If the conversion is successful, but the generated model is wrong, then state what is wrong:

- Model produces wrong results and/or has lesser accuracy.
- Model produces correct results, but it is slower than expected.

### 4. (optional) RNN conversion support

//
// Typical use case, is to infer tensor properties from a graph, before doing
// optimization pass. Nodes modified during optimization pass have to be
// invalidated, to prevent further incorrect optimizations based on wrong shape
// and data type properties.
typedef struct TF_GraphProperties TF_GraphProperties;

// Create GraphProperties. The item must outlive the properties.
TF_CAPI_EXPORT extern TF_GraphProperties* TF_NewGraphProperties(

  TF_DeleteFunction(funcs[0]);
}

TEST_F(CApiExperimentalFunctionTest, EmptyGraphRemoveNonExistentFunction) {
  TF_GraphRemoveFunction(func_graph_, "wrong_name", s_);
  EXPECT_EQ(TF_INVALID_ARGUMENT, TF_GetCode(s_));
  EXPECT_EQ(string("Tried to remove non-existent function 'wrong_name'."),
            string(TF_Message(s_)));
}

TEST_F(CApiExperimentalFunctionTest, GraphRemoveNonExistentFunction) {
  TF_Function* funcs[1];

*   Improved DirectSession::Run() overhead and error checking. Feeding a value
    of the wrong type will now synchronously raise an INVALID_ARGUMENT error
    instead of asynchronously raising an INTERNAL error. Code that depends on
    the (undefined) behavior when feeding a tensor of the wrong type may need to
    be updated.
*   Added unreduced NONE, and reduced MEAN options for losses. Removed

class CustomDeviceOpHandler;
class Device;

// LINT.IfChange
// Note: Keep in sync with exported copy of enum in eager/c_api.h.
enum ContextDevicePlacementPolicy {
  // Running operations with input tensors on the wrong device will fail.
  DEVICE_PLACEMENT_EXPLICIT = 0,
  // Copy the tensor to the right device but log a warning.
  DEVICE_PLACEMENT_WARN = 1,
  // Silently copy the tensor, which has a performance cost since the operation