Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Mar 13, 2021

• [PJRT] Change PJRT to batch stream callbacks that happen at the end of Execute().

  Show description for 588edfc
  

  hawkinsp

  authored and

  tensorflower-gardener

  committed
  588edfc

  Copy full SHA for 588edfc

• Adde split implementation for all other axis(Width/Height/Depth/Batch) in Metal and OpenCL.

  Show description for 0c86f28
  

  roserg

  authored and

  tensorflower-gardener

  committed
  0c86f28

  Copy full SHA for 0c86f28

• compat: Update forward compatibility horizon to 2021-03-13

  Show description for 8ca16a2
  tensorflower-gardener

  committed
  8ca16a2

  Copy full SHA for 8ca16a2

• Update GraphDef version to 704.

  Show description for cece930
  tensorflower-gardener

  committed
  cece930

  Copy full SHA for cece930

• Add block arguments to completely isolate the custom tf op

  Show description for 54fca00
  

  liufengdb

  authored and

  tensorflower-gardener

  committed
  54fca00

  Copy full SHA for 54fca00

• Merge pull request #47677 from ROCmSoftwarePlatform:google_upstream_rocm_build_fix_210309

  Show description for b8a4b26
  tensorflower-gardener

  committed
  b8a4b26

  Copy full SHA for b8a4b26

• Merge pull request #47743 from Intel-tensorflow:upgrading-onednn-to-2.1

  Show description for e0c7549
  tensorflower-gardener

  committed
  e0c7549

  Copy full SHA for e0c7549

• To return tensors from a tf.function, ops.control_dependencies is used. However, where there are hundreds of restored ops, the side effect is adding tens of thousands of edges into the graph, which…

  Show description for 11baa20
  tensorflower-gardener

  committed
  11baa20

  Copy full SHA for 11baa20

• Add support for GPUs in workers for Parameter Server Strategy V2

  Show description for 780b0eb
  tensorflower-gardener

  committed
  780b0eb

  Copy full SHA for 780b0eb

• Add block arguments to completely isolate the custom tf op

  Show description for ad7fc01
  tensorflower-gardener

  committed
  ad7fc01

  Copy full SHA for ad7fc01

• TF API Fixit: Remove the private usage of distribute_utils.is_distributed_variable. DistributedVariable happens to be a subclass of DistributedValue and Variable, both of which are public symbols, …

  Show description for 1a7f3d0
  

  rchao

  authored and

  tensorflower-gardener

  committed
  1a7f3d0

  Copy full SHA for 1a7f3d0

• Merge pull request #47435 from lgeiger:speedup-keras-saved-model-loading

  Show description for 5b505a3
  tensorflower-gardener

  committed
  5b505a3

  Copy full SHA for 5b505a3

• Merge pull request #47598 from annarev:add_bfc_allocator_param

  Show description for 844d24f
  tensorflower-gardener

  committed
  844d24f

  Copy full SHA for 844d24f

• Lower tf.IsFinite to tf.Equal(tf.Subtract(x, x), 0)

  Show description for e89316d
  

  thaink

  authored and

  tensorflower-gardener

  committed
  e89316d

  Copy full SHA for e89316d

• Increase shard count for shard test.

  Show description for dc66f49
  

  aaudiber

  authored and

  tensorflower-gardener

  committed
  dc66f49

  Copy full SHA for dc66f49

• [XLA:Python] Add ProfilerSession bindings.

  Show description for d3a77a5
  

  skye

  authored and

  tensorflower-gardener

  committed
  d3a77a5

  Copy full SHA for d3a77a5

• Fixes reference of constant expression in core/kernels/function_op.cc.

  Show description for 0abdfe6
  tensorflower-gardener

  committed
  0abdfe6

  Copy full SHA for 0abdfe6

• Make data members private. There seems to be no need for them to be public, and public data members are a hard-to-maintain part of the public interface.

  Show description for 2c3f088
  

  tkoeppe

  authored and

  tensorflower-gardener

  committed
  2c3f088

  Copy full SHA for 2c3f088

• Add support for GPUs in workers for Parameter Server Strategy V2

  Show description for 647f495
  

  ishark

  authored and

  tensorflower-gardener

  committed
  647f495

  Copy full SHA for 647f495

• Move SAVED_MODEL_FILENAME_JSON constant into Keras.

  Show description for dd6ceae
  

  k-w-w

  authored and

  tensorflower-gardener

  committed
  dd6ceae

  Copy full SHA for dd6ceae

Commits on Mar 12, 2021

• [lite] Add compatible_with attribute to C API delegate_plugin target.

  Show description for 1271528
  

  fergushenderson

  authored and

  tensorflower-gardener

  committed
  1271528

  Copy full SHA for 1271528

• Copy basic SavedModel utils.

  Show description for 8d4a614
  

  k-w-w

  authored and

  tensorflower-gardener

  committed
  8d4a614

  Copy full SHA for 8d4a614

• Add block arguments to completely isolate the custom tf op

  Show description for a70d683
  

  liufengdb

  authored and

  tensorflower-gardener

  committed
  a70d683

  Copy full SHA for a70d683

• #PRIVATE_TF_API_USAGE_CLEANUP Add register_session_provide to tf.__internal__ for public exposure.

  Show description for fe5cbe5
  

  chenmoneygithub

  authored and

  tensorflower-gardener

  committed
  fe5cbe5

  Copy full SHA for fe5cbe5

• PR #44769: Fix class_names in image_dataset_from_directory

  Show description for 1ea9f94
  tensorflower-gardener

  committed
  1ea9f94

  Copy full SHA for 1ea9f94

• Use old bridge in _TpuCompileMlir op unless MLIR bridge was enabled explicitly

  Show description for 64e374c
  

  smit-hinsu

  authored and

  tensorflower-gardener

  committed
  64e374c

  Copy full SHA for 64e374c

• Add a method to get XEvent metadata ID

  Show description for 0476d2d
  

  Terranlee

  authored and

  tensorflower-gardener

  committed
  0476d2d

  Copy full SHA for 0476d2d

• TPUPartitionedCall: use separate `FunctionLibraryDefinition`s for each

  Show description for 7d232ce
  

  aaroey

  authored and

  tensorflower-gardener

  committed
  7d232ce

  Copy full SHA for 7d232ce

• Transform outside compilation ops generated by the new bridge to old bridge semantics

  Show description for ca211ae
  

  smit-hinsu

  authored and

  tensorflower-gardener

  committed
  ca211ae

  Copy full SHA for ca211ae

• Explicitly delegate learning_rate in LossScaleOptimizer.

  Show description for 31d58c6
  

  reedwm

  authored and

  tensorflower-gardener

  committed
  31d58c6

  Copy full SHA for 31d58c6

• Improve an error message in experimental_split_to_logical_devices.

  Show description for 107665d
  tensorflower-gardener

  committed
  107665d

  Copy full SHA for 107665d

• Generalize tf-rewrite-tpu-embedding-ops pass to handle embedding ops within regions

  Show description for 0b45981
  

  smit-hinsu

  authored and

  tensorflower-gardener

  committed
  0b45981

  Copy full SHA for 0b45981

• [XLA] Fix replay_computation to not link with GPU infeed/outfeed managers directly.

  Show description for 5097e6b
  

  jurahul

  authored and

  tensorflower-gardener

  committed
  5097e6b

  Copy full SHA for 5097e6b

• Do not state deprecated loss scaling APIs will be removed in TF 2.5.

  Show description for a759f01
  

  reedwm

  authored and

  tensorflower-gardener

  committed
  a759f01

  Copy full SHA for a759f01

• Merge pull request #46166 from Intel-tensorflow:quintinwang/quantized_type

  Show description for cbbeb9d
  tensorflower-gardener

  committed
  cbbeb9d

  Copy full SHA for cbbeb9d