*   XLA reduction emitter is deterministic when the environment variable
        `TF_DETERMINISTIC_OPS` is set to "true" or "1". This extends
        deterministic `tf.nn.bias_add` back-prop functionality (and therefore
        also deterministic back-prop of bias-addition in Keras layers) to
        include when XLA JIT compilation is enabled.
    *   Fix problem, when running on a CUDA GPU and when either environment

# This env is only valid when RBE is available on the selected platform, since
# it enables a derived --config setting.  If RBE is not available (i.e. there
# is no --config setting), bazel would fail and quit. This script does a quick
# check This script checks for such errors early
if ! grep "rbe_$TFCI_BAZEL_TARGET_SELECTING_CONFIG_PREFIX" .bazelrc; then
  cat <<EOF

build:sigbuild_remote_cache --remote_cache="https://storage.googleapis.com/tensorflow-devinfra-bazel-cache/manylinux2014" --remote_upload_local_results=false
# Change the value of CACHEBUSTER when upgrading the toolchain, or when testing
# different compilation methods. E.g. for a PR to test a new CUDA version, set
# the CACHEBUSTER to the PR number.
build --action_env=CACHEBUSTER=20220325

    if [[ "$(git rev-parse --abbrev-ref HEAD)" == "pull_branch" ]]; then
        # TF's CI runs 'git fetch origin "pull/PR#/merge:pull_branch"'
        # To get the as-merged branch during the CI tests
        git diff --diff-filter ACMRT --name-only pull_branch^ pull_branch > $BATS_FILE_TMPDIR/changed_files
    else
        # If the branch is not present, then diff against origin/master

# limitations under the License.
# ==============================================================================
# Sourcing this enables local disk cache
if [[ $(uname -s) == "Darwin" ]]; then
  echo "Please note that using disk cache on macOS is not recommended because the"
  echo "cache can end up being pretty big and make the build process inefficient."
fi

        "//tensorflow/core:test_main",
    ],
)

tf_cc_test(
    name = "parallel_device_remote_test",
    srcs = ["parallel_device_remote_test.cc"],
    # TODO(b/136478427): Enable global heap checking when servers shut down
    # cleanly.
    args = ["--heap_check="],
    deps = [
        ":parallel_device",
        ":parallel_device_testlib",
        "//tensorflow/c:c_api",

  * If you are an individual writing original source code and you're sure you own the intellectual property, then you'll need to sign an [individual CLA](https://code.google.com/legal/individual-cla-v1.0.html).
  * If you work for a company that wants to allow you to contribute your work, then you'll need to sign a [corporate CLA](https://code.google.com/legal/corporate-cla-v1.0.html).

#
#   Bazel will cache your builds in tensorflow/build_output/cache,
#   and will also try using public build cache results to speed up
#   your builds. This usually saves a lot of time, especially when
#   re-running tests. However, note that:
#
#    - New environments like new CUDA versions, changes to manylinux,
#      compilers, etc. can cause undefined behavior such as build failures
#      or tests passing incorrectly.

- 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

#
# The cascade of load() statements and tf_workspace?() calls works around the
# restriction that load() statements need to be at the top of .bzl files.
# E.g. we can not retrieve a new repository with http_archive and then load()
# a macro from that repository in the same file.
load("@//tensorflow:workspace3.bzl", "tf_workspace3")

tf_workspace3()

load("@//tensorflow:workspace2.bzl", "tf_workspace2")

tf_workspace2()