virtual Status NumElements(int64_t* num_elements) const = 0;
  // Returns size of specified dimension
  //
  // -1 indicates an unknown axis length; this is unreachable for most standard
  // ImmediateExecutionTensorHandles, but comes up for example when computing
  // the shape of a parallel tensor with component shapes differing across
  // devices.
  virtual Status Dim(int dim_index, int64_t* dim) const = 0;

    -   Uses `pycpp.sh`
-   Presubmit jobs (Run on every GitHub PR)
    -   Uses `pycpp.sh`, `code_check_changed_files.sh`

These "env" files match up with an environment matrix that roughly covers:

-   Different Python versions
-   Linux, MacOS, and Windows machines (these pool definitions are internal)
-   x86 and arm64
-   CPU-only, or with NVIDIA CUDA support (Linux only), or with TPUs

}

void TF_ImportGraphDefResultsReturnOperations(TF_ImportGraphDefResults* results,
                                              int* num_opers,
                                              TF_Operation*** opers) {
  *num_opers = results->return_nodes.size();
  *opers = results->return_nodes.data();
}

void TF_ImportGraphDefResultsMissingUnusedInputMappings(

 # define _strong_alias(name, aliasname) \
-  extern __typeof (name) aliasname __attribute__ ((alias (#name)));
+  extern __typeof (name) aliasname __attribute__ ((alias (#name))) \
+    __attribute_copy__ (name);
 
 /* This comes between the return type and function name in
    a function definition to make that definition weak.  */
@@ -125,14 +131,16 @@
    If weak aliases are not available, this defines a strong alias.  */

      if (existing != accumulated_gradients_.end()) {
        // This is a somewhat odd case to be in, since it means we have two
        // operations which supposedly both created the same Tensor. It comes up
        // in recompute_grad, where the gradients have the same value. However,
        // only the original gradient is connected to everything else, so we
        // should still use that.

    }
    return out;
  }

  void Define(int num_opers, const std::vector<TF_Operation*>& opers,
              const std::vector<TF_Operation*>& inputs,
              const std::vector<TF_Operation*>& outputs,
              const std::vector<string>& output_names,
              bool expect_failure = false) {
    DefineT(num_opers, opers, ToOutput(inputs), ToOutput(outputs), output_names,
            expect_failure);

vulnerabilities to the vendor of the affected hardware accelerator.

## Reporting vulnerabilities

### Vulnerabilities in TensorFlow

This document covers different use cases for TensorFlow together with comments
whether these uses were recommended or considered safe, or where we recommend
some form of isolation when dealing with untrusted data. As a result, this

  TF_Output inputs[] = {{feed, 0}};
  TF_Output outputs[] = {{neg, 0}};
  *func = TF_GraphToFunction(func_graph.get(), name, append_hash, -1,
                             /*opers=*/nullptr, 1, inputs, 1, outputs,
                             /*output_names=*/nullptr,
                             /*opts=*/nullptr, description, s.get());
  ASSERT_EQ(TF_OK, TF_GetCode(s.get())) << TF_Message(s.get());

    and keyword arguments when calling the layer.
*   Functional models now get constructed if *any* tensor in a layer call's
    arguments/keyword arguments comes from a keras input. Previously the
    functional api would only work if all of the elements in the first argument
    to the layer came from a keras input.

// Performs various checks.
Status ComputeBodyNodes(
    const TF_Graph* fn_body, const char* fn_name, int num_opers,
    const TF_Operation* const* opers,
    const std::unordered_map<const Node*, std::vector<int>>& input_nodes,
    std::vector<const Node*>* body_nodes)
    TF_EXCLUSIVE_LOCKS_REQUIRED(fn_body->mu) {
  if (num_opers == -1) {
    for (const Node* node : fn_body->graph.op_nodes()) {
      const auto& iter = input_nodes.find(node);