Merge branch 'tensorflow:master' into matthew-olson-intel/csr_matmul

.bazelrc

@@ -255,6 +255,14 @@ build:cuda_clang_official --action_env=CLANG_CUDA_COMPILER_PATH="/usr/lib/llvm-1
 build:cuda_clang_official --action_env=LD_LIBRARY_PATH="/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64"
 build:cuda_clang_official --crosstool_top="@sigbuild-r2.16-clang_config_cuda//crosstool:toolchain"
 
+# Build with nvcc for CUDA and clang for host
+build:nvcc_clang --config=cuda
+# Unfortunately, cuda_configure.bzl demands this for using nvcc + clang
+build:nvcc_clang --action_env=TF_CUDA_CLANG="1"
+build:nvcc_clang --action_env=TF_NVCC_CLANG="1"
+build:nvcc_clang --@local_config_cuda//:cuda_compiler=nvcc
+
+
 # Debug config
 build:dbg -c dbg
 # Only include debug info for files under tensorflow/, excluding kernels, to
@@ -527,8 +535,8 @@ build:rbe_linux_cuda --repo_env=TF_NCCL_CONFIG_REPO="@sigbuild-r2.16-clang_confi
 test:rbe_linux_cuda --test_env=LD_LIBRARY_PATH="/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64"
 
 build:rbe_linux_cuda_nvcc --config=rbe_linux_cuda
+build:rbe_linux_cuda_nvcc --config=nvcc_clang
 build:rbe_linux_cuda_nvcc --repo_env TF_NCCL_USE_STUB=1
-build:rbe_linux_cuda_nvcc --action_env=TF_NVCC_CLANG="1"
 
 # TODO(kanglan): Remove rbe_win and rbe_win_py3* after b/289091160 is fixed
 build:rbe_win --config=rbe_base
@@ -577,6 +585,7 @@ build:elinux_armhf --copt -mfp16-format=ieee
 
 # Load rc file written by ./configure.
 try-import %workspace%/.tf_configure.bazelrc
+try-import %workspace%/xla_configure.bazelrc
 
 # Load rc file with user-specific options.
 try-import %workspace%/.bazelrc.user
@@ -777,28 +786,38 @@ test:linux_cuda_pycpp_test_filters --build_tag_filters=-no_oss,-oss_excluded,-os
 test:linux_cuda_pycpp_test_filters --test_lang_filters=cc,py --test_size_filters=small,medium
 test:linux_cuda_pycpp_test --config=linux_cuda_pycpp_test_filters -- //tensorflow/... -//tensorflow/python/integration_testing/... -//tensorflow/compiler/tf2tensorrt/... -//tensorflow/core/tpu/... -//tensorflow/lite/... -//tensorflow/tools/toolchains/...
 # ARM64 PYCPP
-test:linux_arm64_pycpp_test_filters --test_tag_filters=-no_oss,-no_aarch64,-oss_excluded,-oss_serial,-gpu,-tpu,-benchmark-test,-v1only
-test:linux_arm64_pycpp_test_filters --build_tag_filters=-no_oss,-no_aarch64,-oss_excluded,-oss_serial,-gpu,-tpu,-benchmark-test,-v1only
-test:linux_arm64_pycpp_test_filters --test_lang_filters=cc,py --test_size_filters=small,medium --flaky_test_attempts=3
+# In Linux Arm64 presubmit/continuous build, we cross-compile the binaries on
+# Linux x86 so that we can use RBE. Since tests still need to run on the single
+# host Arm64 machine, the build becomes too slow (~30 min) to be a presubmit.
+# For testing purposes, we want to see the runtime performance of an
+# experimental job that is build-only, i.e, we only build the test targets and
+# do not run them. By prefixing the configs with "build", we can run both
+# `bazel build` and `bazel test` commands with the same config as test configs
+# inherit from build.
+build:linux_arm64_pycpp_test_filters --test_tag_filters=-no_oss,-no_aarch64,-oss_excluded,-oss_serial,-gpu,-tpu,-benchmark-test,-v1only
+build:linux_arm64_pycpp_test_filters --build_tag_filters=-no_oss,-no_aarch64,-oss_excluded,-oss_serial,-gpu,-tpu,-benchmark-test,-v1only
+build:linux_arm64_pycpp_test_filters --test_lang_filters=cc,py --test_size_filters=small,medium --flaky_test_attempts=3
 # TODO(michaelhudgins): Why do we need to specifically omit go and java here? 
-test:linux_arm64_pycpp_test --config=linux_arm64_pycpp_test_filters -- //tensorflow/... -//tensorflow/python/integration_testing/... -//tensorflow/compiler/tf2tensorrt/... -//tensorflow/core/tpu/... -//tensorflow/lite/... -//tensorflow/tools/toolchains/... -//tensorflow/go/... -//tensorflow/java/... -//tensorflow/core/grappler/optimizers:auto_mixed_precision_test_cpu -//tensorflow/core/grappler/optimizers:remapper_test_cpu -//tensorflow/core/kernels/image:resize_bicubic_op_test -//tensorflow/compiler/mlir/tfr/examples/customization:test_ops_test -//tensorflow/compiler/mlir/tfr/examples/mnist:mnist_ops_test -//tensorflow/compiler/mlir/tfr/examples/pad:pad_ops_test -//tensorflow/python/tools:aot_compiled_test
+build:linux_arm64_pycpp_test --config=linux_arm64_pycpp_test_filters -- //tensorflow/... -//tensorflow/python/integration_testing/... -//tensorflow/compiler/tf2tensorrt/... -//tensorflow/core/tpu/... -//tensorflow/lite/... -//tensorflow/tools/toolchains/... -//tensorflow/go/... -//tensorflow/java/... -//tensorflow/core/grappler/optimizers:auto_mixed_precision_test_cpu -//tensorflow/core/grappler/optimizers:remapper_test_cpu -//tensorflow/core/kernels/image:resize_bicubic_op_test -//tensorflow/compiler/mlir/tfr/examples/customization:test_ops_test -//tensorflow/compiler/mlir/tfr/examples/mnist:mnist_ops_test -//tensorflow/compiler/mlir/tfr/examples/pad:pad_ops_test -//tensorflow/python/tools:aot_compiled_test
 # CROSS-COMPILE ARM64 PYCPP
-test:cross_compile_linux_arm64_pycpp_test --config=linux_arm64_pycpp_test
+build:cross_compile_linux_arm64_pycpp_test --config=linux_arm64_pycpp_test
 # Tests that fail only when cross-compiled
-test:cross_compile_linux_arm64_pycpp_test -//tensorflow/compiler/mlir/quantization/stablehlo:convert_tf_quant_to_mhlo_int_test
+build:cross_compile_linux_arm64_pycpp_test -//tensorflow/compiler/mlir/quantization/stablehlo:convert_tf_quant_to_mhlo_int_test
 # MACOS ARM64 PYCPP
 test:macos_arm64_pycpp_test_filters --test_tag_filters=-no_oss,-oss_excluded,-oss_serial,-no_oss_py39,-no_oss_py310,-nomac,-no_mac,-mac_excluded,-v1only,-gpu,-tpu,-benchmark-test,-no_mac_arm64,-no_aarch64
 test:macos_arm64_pycpp_test_filters --build_tag_filters=-no_oss,-oss_excluded,-oss_serial,-no_oss_py39,-no_oss_py310,-nomac,-no_mac,-mac_excluded,-v1only,-gpu,-tpu,-benchmark-test,-no_mac_arm64,-no_aarch64
 test:macos_arm64_pycpp_test_filters --test_lang_filters=cc,py --test_size_filters=small,medium
 test:macos_arm64_pycpp_test --config=macos_arm64_pycpp_test_filters -- //tensorflow/... -//tensorflow/python/integration_testing/... -//tensorflow/compiler/tf2tensorrt/... -//tensorflow/core/tpu/... -//tensorflow/lite/... -//tensorflow/tools/toolchains/... -//tensorflow/go/... -//tensorflow/java/... -//tensorflow/compiler/aot/... -//tensorflow/core/kernels/image:resize_bicubic_op_test
 # MACOS X86 PYCPP
-test:macos_x86_pycpp_test_filters --test_tag_filters=-no_oss,-oss_excluded,-oss_serial,-no_oss_py38,-no_oss_py39,-no_oss_py310,-nomac,-no_mac,-mac_excluded,-v1only,-gpu,-tpu,-benchmark-test
-test:macos_x86_pycpp_test_filters --build_tag_filters=-no_oss,-oss_excluded,-oss_serial,-no_oss_py38,-no_oss_py39,-no_oss_py310,-nomac,-no_mac,-mac_excluded,-v1only,-gpu,-tpu,-benchmark-test
-test:macos_x86_pycpp_test_filters --keep_going --test_lang_filters=cc,py --test_size_filters=small,medium
-test:macos_x86_pycpp_test --config=macos_x86_pycpp_test_filters -- //tensorflow/... -//tensorflow/compiler/tf2tensorrt/... -//tensorflow/core/tpu/... -//tensorflow/go/... -//tensorflow/java/... -//tensorflow/python/integration_testing/... -//tensorflow/tools/toolchains/... -//tensorflow/lite/... -//tensorflow/compiler/aot/...
+# These are defined as build configs so that we can run a build only job. See
+# the note under "ARM64 PYCPP" for more details.
+build:macos_x86_pycpp_test_filters --test_tag_filters=-no_oss,-oss_excluded,-oss_serial,-no_oss_py38,-no_oss_py39,-no_oss_py310,-nomac,-no_mac,-mac_excluded,-v1only,-gpu,-tpu,-benchmark-test
+build:macos_x86_pycpp_test_filters --build_tag_filters=-no_oss,-oss_excluded,-oss_serial,-no_oss_py38,-no_oss_py39,-no_oss_py310,-nomac,-no_mac,-mac_excluded,-v1only,-gpu,-tpu,-benchmark-test
+build:macos_x86_pycpp_test_filters --keep_going --test_lang_filters=cc,py --test_size_filters=small,medium
+build:macos_x86_pycpp_test --config=macos_x86_pycpp_test_filters -- //tensorflow/... -//tensorflow/compiler/tf2tensorrt/... -//tensorflow/core/tpu/... -//tensorflow/go/... -//tensorflow/java/... -//tensorflow/python/integration_testing/... -//tensorflow/tools/toolchains/... -//tensorflow/lite/... -//tensorflow/compiler/aot/...
 # CROSS-COMPILE MACOS X86 PYCPP
-test:cross_compile_macos_x86_pycpp_test --config=macos_x86_pycpp_test
-test:cross_compile_macos_x86_pycpp_test -//tensorflow/core/kernels:quantized_conv_ops_test -//tensorflow/core/kernels:quantized_matmul_op_test -//tensorflow/python/ops:quantized_conv_ops_test -//tensorflow/tools/graph_transforms:transforms_test -//tensorflow/python/tools:aot_compiled_test
+build:cross_compile_macos_x86_pycpp_test --config=macos_x86_pycpp_test
+build:cross_compile_macos_x86_pycpp_test -//tensorflow/core/kernels:quantized_conv_ops_test -//tensorflow/core/kernels:quantized_matmul_op_test -//tensorflow/python/ops:quantized_conv_ops_test -//tensorflow/tools/graph_transforms:transforms_test -//tensorflow/python/tools:aot_compiled_test
 # END TF TEST SUITE OPTIONS
 
 # START CROSS-COMPILE CONFIGS
@@ -855,8 +874,12 @@ build:cross_compile_macos_x86 --platform_mappings=tensorflow/tools/toolchains/cr
 # RBE cross-compile configs for Darwin x86
 build:rbe_cross_compile_macos_x86 --config=cross_compile_macos_x86
 build:rbe_cross_compile_macos_x86 --config=rbe_cross_compile_base
+build:rbe_cross_compile_macos_x86 --bes_upload_mode=nowait_for_upload_complete
 test:rbe_cross_compile_macos_x86 --config=rbe_cross_compile_base
 # Increase the test timeout as tests often take longer on mac.
 test:rbe_cross_compile_macos_x86 --test_timeout=300,450,1200,3600
+# Limit jobs to 100 to avoid running into "out of memory" issues (b/316266643)
+build:rbe_cross_compile_macos_x86 --jobs=100
+test:rbe_cross_compile_macos_x86 --jobs=100
 # END MACOS CROSS-COMPILE CONFIGS
 # END CROSS-COMPILE CONFIGS

.github/workflows/arm-cd.yml

@@ -57,12 +57,12 @@ jobs:
         run: find /home/ubuntu/actions-runner/_work/tensorflow/tensorflow/. -name . -o -prune -exec sudo rm -rf -- {} + || true
       - name: Checkout repository for nightly (skipped for releases)
         if: ${{ github.event_name == 'schedule' }}
-        uses: actions/checkout@755da8c3cf115ac066823e79a1e1788f8940201b # v3.2.0
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
         with:
           ref: 'nightly'
       - name: Checkout repository for releases (skipped for nightly)
         if: ${{ github.event_name == 'push' }}
-        uses: actions/checkout@755da8c3cf115ac066823e79a1e1788f8940201b # v3.2.0
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
       - name: Build and test pip wheel
         shell: bash
         run: |

.github/workflows/arm-ci-extended-cpp.yml

@@ -50,12 +50,12 @@ jobs:
         run: find /home/ubuntu/actions-runner/_work/tensorflow/tensorflow/. -name . -o -prune -exec sudo rm -rf -- {} + || true
       - name: Checkout repository for nightly (skipped for releases)
         if: ${{ github.event_name == 'schedule' }}
-        uses: actions/checkout@755da8c3cf115ac066823e79a1e1788f8940201b # v3.2.0
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
         with:
           ref: 'nightly'
       - name: Checkout repository
         if: ${{ github.event_name == 'push' }}
-        uses: actions/checkout@755da8c3cf115ac066823e79a1e1788f8940201b # v3.2.0
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
       - name: Build binary and run C++ tests
         shell: bash
         run: |

.github/workflows/arm-ci-extended.yml

@@ -51,12 +51,12 @@ jobs:
         run: find /home/ubuntu/actions-runner/_work/tensorflow/tensorflow/. -name . -o -prune -exec sudo rm -rf -- {} + || true
       - name: Checkout repository for nightly (skipped for releases)
         if: ${{ github.event_name == 'schedule' }}
-        uses: actions/checkout@755da8c3cf115ac066823e79a1e1788f8940201b # v3.2.0
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
         with:
           ref: 'nightly'
       - name: Checkout repository
         if: ${{ github.event_name == 'push' }}
-        uses: actions/checkout@755da8c3cf115ac066823e79a1e1788f8940201b # v3.2.0
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
       - name: Build binary and run python tests on nightly for all python versions
         shell: bash
         run: |

.github/workflows/arm-ci.yml

@@ -47,7 +47,7 @@ jobs:
         shell: bash
         run: find /home/ubuntu/actions-runner/_work/tensorflow/tensorflow/. -name . -o -prune -exec sudo rm -rf -- {} + || true
       - name: Checkout repository
-        uses: actions/checkout@755da8c3cf115ac066823e79a1e1788f8940201b # v3.2.0
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
       - name: Build binary and run python tests
         shell: bash
         run: |

.github/workflows/osv-scanner-scheduled.yml

@@ -28,7 +28,7 @@ permissions:
 jobs:
   scan-scheduled:
     if: github.repository == 'tensorflow/tensorflow'
-    uses: "google/osv-scanner/.github/workflows/osv-scanner-reusable.yml@main"
+    uses: "google/osv-scanner-action/.github/workflows/osv-scanner-reusable.yml@v1.6.2-beta1"
     with:
       scan-args: |-
         --lockfile=requirements.txt:./requirements_lock_3_9.txt

.github/workflows/update-rbe.yml

@@ -106,13 +106,13 @@ jobs:
         map sigbuild-r2.14-clang-python3.10 2.14-python3.10
         map sigbuild-r2.14-clang-python3.11 2.14-python3.11
         # TF 2.16
-        map sigbuild-r2.16 2.16-python3.9
+        map sigbuild-r2.16 2.16-python3.11
         map sigbuild-r2.16-python3.9 2.16-python3.9
         map sigbuild-r2.16-python3.10 2.16-python3.10
         map sigbuild-r2.16-python3.11 2.16-python3.11
         map sigbuild-r2.16-python3.12 2.16-python3.12
         # TF 2.16 + Clang (containers are the same, but env vars in configs.bzl are different)
-        map sigbuild-r2.16-clang 2.16-python3.9
+        map sigbuild-r2.16-clang 2.16-python3.11
         map sigbuild-r2.16-clang-python3.9 2.16-python3.9
         map sigbuild-r2.16-clang-python3.10 2.16-python3.10
         map sigbuild-r2.16-clang-python3.11 2.16-python3.11

.gitignore

@@ -3,6 +3,7 @@
 node_modules
 /.bazelrc.user
 /.tf_configure.bazelrc
+/xla_configure.bazelrc
 /bazel-*
 /bazel_pip
 /tools/python_bin_path.sh

RELEASE.md

@@ -1,4 +1,4 @@
-# Release 2.16.0
+# Release 2.17.0
 
 ## TensorFlow
 
@@ -9,11 +9,31 @@
 * <DOCUMENT BREAKING CHANGES HERE>
 * <THIS SECTION SHOULD CONTAIN API, ABI AND BEHAVIORAL BREAKING CHANGES>
 
-* `tf.summary.trace_on` now takes a `profiler_outdir` argument. This must be set
-  if `profiler` arg is set to `True`.
-    * `tf.summary.trace_export`'s `profiler_outdir` arg is now a no-op. Enabling
-      the profiler now requires setting `profiler_outdir` in `trace_on`.
+### Known Caveats
 
+* <CAVEATS REGARDING THE RELEASE (BUT NOT BREAKING CHANGES).>
+* <ADDING/BUMPING DEPENDENCIES SHOULD GO HERE>
+* <KNOWN LACK OF SUPPORT ON SOME PLATFORM, SHOULD GO HERE>
+
+### Major Features and Improvements
+
+*   <INSERT MAJOR FEATURE HERE, USING MARKDOWN SYNTAX>
+*   <IF RELEASE CONTAINS MULTIPLE FEATURES FROM SAME AREA, GROUP THEM TOGETHER>
+
+### Bug Fixes and Other Changes
+
+* <SIMILAR TO ABOVE SECTION, BUT FOR OTHER IMPORTANT CHANGES / BUG FIXES>
+* <IF A CHANGE CLOSES A GITHUB ISSUE, IT SHOULD BE DOCUMENTED HERE>
+* <NOTES SHOULD BE GROUPED PER AREA>
+
+## Keras
+
+<INSERT SMALL BLURB ABOUT RELEASE FOCUS AREA AND POTENTIAL TOOLCHAIN CHANGES>
+
+### Breaking Changes
+
+* <DOCUMENT BREAKING CHANGES HERE>
+* <THIS SECTION SHOULD CONTAIN API, ABI AND BEHAVIORAL BREAKING CHANGES>
 
 ### Known Caveats
 
@@ -26,6 +46,101 @@
 *   <INSERT MAJOR FEATURE HERE, USING MARKDOWN SYNTAX>
 *   <IF RELEASE CONTAINS MULTIPLE FEATURES FROM SAME AREA, GROUP THEM TOGETHER>
 
+### Bug Fixes and Other Changes
+
+* <SIMILAR TO ABOVE SECTION, BUT FOR OTHER IMPORTANT CHANGES / BUG FIXES>
+* <IF A CHANGE CLOSES A GITHUB ISSUE, IT SHOULD BE DOCUMENTED HERE>
+* <NOTES SHOULD BE GROUPED PER AREA>
+
+* `tf.lite`
+    * Quantization for `FullyConnected` layer is switched from per-tensor to
+      per-channel scales for dynamic range quantization use case (`float32`
+      inputs / outputs and `int8` weights). The change enables new quantization
+      schema globally in the converter and inference engine. The new behaviour
+      can be disabled via experimental
+      flag `converter._experimental_disable_per_channel_quantization_for_dense_layers = True`.
+
+## Thanks to our Contributors
+
+This release contains contributions from many people at Google, as well as:
+
+<INSERT>, <NAME>, <HERE>, <USING>, <GITHUB>, <HANDLE>
+
+# Release 2.16.0
+
+## TensorFlow
+
+<INSERT SMALL BLURB ABOUT RELEASE FOCUS AREA AND POTENTIAL TOOLCHAIN CHANGES>
+
+*   TensorFlow Windows Build:
+
+    *   Clang is now the default compiler to build TensorFlow CPU wheels on the
+        Windows Platform starting with this release. The currently supported
+        version is LLVM/clang 17. The official Wheels-published on PyPI will be
+        based on Clang; however, users retain the option to build wheels using
+        the MSVC compiler following the steps mentioned in
+        https://www.tensorflow.org/install/source_windows as has been the case
+        before
+
+### Breaking Changes
+
+*   <DOCUMENT BREAKING CHANGES HERE>
+*   <THIS SECTION SHOULD CONTAIN API, ABI AND BEHAVIORAL BREAKING CHANGES>
+
+*   `tf.summary.trace_on` now takes a `profiler_outdir` argument. This must be
+    set if `profiler` arg is set to `True`.
+
+    *   `tf.summary.trace_export`'s `profiler_outdir` arg is now a no-op.
+        Enabling the profiler now requires setting `profiler_outdir` in
+        `trace_on`.
+
+*   `tf.estimator`
+
+    *   The tf.estimator API is removed.
+
+*   Keras 3.0 will be the default Keras version. You may need to update your
+    script to use Keras 3.0.
+
+*   Please refer to the new Keras documentation for Keras 3.0
+    (https://keras.io/keras_3).
+
+*   To continue using Keras 2.0, do the following.
+
+*   1.  Install tf-keras via pip install tf-keras~=2.16
+
+    1.  To switch tf.keras to use Keras 2 (tf-keras), set the environment
+        variable TF_USE_LEGACY_KERAS=1 directly or in your python program by
+        import os;os.environ["TF_USE_LEGACY_KERAS"]=1. Please note that this
+        will set it for all packages in your Python runtime program
+
+*   1.  Change import of keras from tensorflow as follows
+*   import tensorflow.keras as keras and import keras to import tf_keras as
+    keras
+* **Apple Silicon users:** If you previously installed TensorFlow using
+    `pip install tensorflow-macos`, please update your installation method. Use
+    `pip install tensorflow` from now on. Starting with TF 2.17, the
+    `tensorflow-macos` package will no longer receive updates.
+
+### Known Caveats
+
+* <CAVEATS REGARDING THE RELEASE (BUT NOT BREAKING CHANGES).>
+* <ADDING/BUMPING DEPENDENCIES SHOULD GO HERE>
+* <KNOWN LACK OF SUPPORT ON SOME PLATFORM, SHOULD GO HERE>
+
+*  Full aarch64 Linux and Arm64 macOS wheels are now published to the
+  `tensorflow` pypi repository and no longer redirect to a separate package.
+
+### Major Features and Improvements
+
+*   <INSERT MAJOR FEATURE HERE, USING MARKDOWN SYNTAX>
+*   <IF RELEASE CONTAINS MULTIPLE FEATURES FROM SAME AREA, GROUP THEM TOGETHER>
+
+*  Support for Python 3.12 has been added.
+*  [tensorflow-tpu](https://pypi.org/project/tensorflow-tpu/) package is now
+   available for easier TPU based installs.
+*  TensorFlow pip packages are now built with CUDA 12.3 and cuDNN 8.9.7
+
+
 ### Bug Fixes and Other Changes
 
 * <SIMILAR TO ABOVE SECTION, BUT FOR OTHER IMPORTANT CHANGES / BUG FIXES>
@@ -54,6 +169,21 @@
     * Added `experimental_skip_slot_variables` (a boolean option) to skip
     restoring of optimizer slot variables in a checkpoint.
 
+*   `tf.saved_model.SaveOptions`
+
+    * `SaveOptions` now takes a new argument called
+      `experimental_debug_stripper`. When enabled, this strips the debug nodes
+      from both the node defs and the function defs of the graph. Note that
+      this currently only strips the `Assert` nodes from the graph and
+      converts them into `NoOp`s instead.
+
+*   `tf.data`
+
+    * `tf.data` now has an `autotune_options.initial_parallelism` option to
+      control the initial parallelism setting used by autotune before the data
+      pipeline has started running. The default is 16. A lower value reduces
+      initial memory usage, while a higher value improves startup time.
+
 ## Keras
 
 *  `keras.layers.experimental.DynamicEmbedding`