`ModelServer` collates all computation graphs exposed to it (from multiple
`SavedModel`) and executes them in parallel on available executors. Running
TensorFlow in a multitenant design mixes the risks described above with the
inherent ones from multitenant configurations. The primary areas of concern are
tenant isolation, resource allocation, model sharing and hardware attacks.

### Tenant isolation

  if (av_tensor == nullptr) {
    TF_ASSIGN_OR_RETURN(
        *tensor, MakeTensorFromPjRtBuffer(tensor->dtype(), tensor->shape(),
                                          std::move(buffer)));
  } else {
    av_tensor->SetBuffer(std::move(buffer));
  }
  return absl::OkStatus();
}

absl::StatusOr<xla::PjRtCApiClient*> GetPjRtCApiClient(
    const DeviceType& device_type) {

# ==============================================================================

name: Trusted Partner PR
on:
  pull_request_target:

permissions:
  contents: read

jobs:
  assign-partner-tags:
    runs-on: ubuntu-latest
    permissions:
      # Needed to attach tags into the PR
      issues: write
      contents: write
      pull-requests: write
    if: |

    throw "Error adding labels to PR";
  }
  if (assignees.length > 0) {
    const resp_assign = await github.rest.issues.addAssignees({
      issue_number: context.issue.number,
      owner: context.repo.owner,
      repo: context.repo.repo,
      assignees: assignees
    });
    if (resp_assign.status >= 400) {
      console.log(resp_assign);
      throw "Error adding assignee to PR";
    }
  }

multiple computational devices, including multicore CPUs, general purpose GPUs, and custom-designed ASICs known as Tensor Processing Units (TPUs). This architecture gives flexibility to the application developer, whereas in previous “parameter server” designs the management of shared state is built into the system, TensorFlow enables developers to experiment with novel optimizations and training algorithms. TensorFlow supports a variety of applications, with a focus on training and inference on deep neural...

  tensorflow::Tensor tensor(allocator.get(), DT_FLOAT, {1});
  TF_ASSERT_OK_AND_ASSIGN(
      auto pjrt_client,
      xla::GetTfrtCpuClient(/*asynchronous=*/true, /*cpu_device_count=*/1));
  std::vector<int32_t> data(1, 0);
  xla::Shape shape = xla::ShapeUtil::MakeShape(xla::S32, {1});
  TF_ASSERT_OK_AND_ASSIGN(
      auto buffer,
      pjrt_client->BufferFromHostBuffer(

  // Assigns a new value to the variable.
  void Assign(TFE_Context* context, TFE_TensorHandle* value, TF_Status* status);
  // Adds `value` to the existing value of the variable.
  void AssignAdd(TFE_Context* context, TFE_TensorHandle* value,
                 TF_Status* status);

 private:
  // Helper for running any single-argument assignment ops (Assign, AssignAdd,
  // AssignSub, ...).

    console.log(pr_resp);
    throw `PR:{pr_number} needs to be valid and closed (merged)`;
  }
  const pr_title = pr_resp.data.title;
  // Assign to PR owner and reviewers
  const assignees = pr_resp.data.assignees.concat(pr_resp.data.requested_reviewers);
  let assignee_logins = assignees.map(x => x.login);
  assignee_logins.push(pr_resp.data.user.login);