AR_IMAGE_PATH="us-central1-docker.pkg.dev/tensorflow-sigs/tensorflow/build-arm64"

# Build for both JAX and TF usage.  We do these in one place because they share
# almost all of the same cache layers
export DOCKER_BUILDKIT=1
for target in jax tf; do
  IMAGE="gcr.io/tensorflow-sigs/build-arm64:$target-$TAG"
  AR_IMAGE="$AR_IMAGE_PATH:$target-$TAG"
  docker pull "$IMAGE" || true

"usability.observability.status" is the feature, and "exist-by-default" is the scenario in the first case, and the second case has no scenario.

The hierarchical nature of feature labels allows us to aggregate data about related feature sets.  To provide for consistent reporting and aggregation, we have defined the top two layers of hierarchy, and ask that you place your feature under these headings.  For more detail on the purpose of each heading, see Top-Level Feature Headings below.  If you feel that...

 *   Added `jit_compile` as a settable property to `tf.keras.Model`.
 *   Added `synchronized` optional parameter to `layers.BatchNormalization`.

  // CHECK-SAME: %[[CONV2]]
  // CHECK-SAME: data_format = "NHWC"

  // ------------------------------------------------------------------------ //
  // Add results of convolution layers #1 and #2.
  // ------------------------------------------------------------------------ //

  %14 = "tf.AddV2"(%10, %12) : (tensor<?x256x56x56xf32>, tensor<?x256x56x56xf32>) -> tensor<?x256x56x56xf32>

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",

  },
  "layered_runtime": {
      "layers": [
          {
            "name": "global config",