)

# buildifier: disable=same-origin-load
load("@rules_python//python:repositories.bzl", "py_repositories")

py_repositories()

load("@rules_python//python:repositories.bzl", "python_register_multi_toolchains")  # buildifier: disable=same-origin-load
load("@rules_python//python/pip_install:repositories.bzl", "pip_install_dependencies")

default_python_version = "3.10"

load(
    "//:updater_config_repository.bzl",

# despite lacking Nvidia CUDA support.
TFCI_BUILD_PIP_PACKAGE_ARGS="--repo_env=WHEEL_NAME=tensorflow"
TFCI_DOCKER_ENABLE=1
TFCI_DOCKER_IMAGE=gcr.io/tensorflow-sigs/build-arm64:tf-2-16-multi-python
TFCI_DOCKER_PULL_ENABLE=1
TFCI_DOCKER_REBUILD_ARGS="--target=tf ci/official/containers/linux_arm64"
TFCI_INDEX_HTML_ENABLE=1
TFCI_LIB_SUFFIX="-cpu-linux-arm64"
TFCI_OUTPUT_DIR=build_output
TFCI_WHL_AUDIT_ENABLE=1

   actually using the new version with Tensorflow.

2) In the `WORKSPACE` file, add the new version to the `python_versions`
   parameter of the `python_register_multi_toolchains` function.

3) In the `BUILD.bazel` file, add a load statement for the new version, e.g.

   ```
      load("@python//3.11:defs.bzl",
           compile_pip_requirements_3_11 = "compile_pip_requirements")

    "python_repository",
)

python_repository(name = "python_version_repo")

load("@python_version_repo//:py_version.bzl", "TF_PYTHON_VERSION")

# Register multi toolchains
load("@rules_python//python:repositories.bzl", "python_register_toolchains")  # buildifier: disable=same-origin-load

python_register_toolchains(
    name = "python",
    ignore_root_user_error = True,

stated above is already equivalent to code execution. In any case, the
recommendation is not to serve models using Eager mode since it also has
performance limitations.

## Multi-Tenant environments

It is possible to run multiple TensorFlow models in parallel. For example,
`ModelServer` collates all computation graphs exposed to it (from multiple

        *   Renamed `"binary"` output mode to `"multi_hot"` for
            `CategoryEncoding`, `StringLookup`, `IntegerLookup`, and
            `TextVectorization`. Multi-hot encoding will no longer automatically
            uprank rank 1 inputs, so these layers can now multi-hot encode
            unbatched multi-dimensional samples.
        *   Added a new output mode `"one_hot"` for `CategoryEncoding`,

    size = "medium",
    srcs = [
        "c_api_debug_test.cc",
        "c_api_test.cc",
    ],
    tags = [
        "no_cuda_asan",  # TODO(b/181771536)
        "guitar",
        # "multi_gpu",  b/180748118
    ],
    deps = [
        ":c_api",
        ":c_api_experimental",
        ":c_api_internal",
        ":c_api_test_util",
        ":tfe_op_internal",

build:ios_arm64e --cpu=ios_arm64e
build:ios_sim_arm64 --config=ios
build:ios_sim_arm64 --cpu=ios_sim_arm64
build:ios_x86_64 --config=ios
build:ios_x86_64 --cpu=ios_x86_64
build:ios_fat --config=ios
build:ios_fat --ios_multi_cpus=armv7,arm64,i386,x86_64

# Config to use a mostly-static build and disable modular op registration
# support (this will revert to loading TensorFlow with RTLD_GLOBAL in Python).