[pull] master from tensorflow:master - #8796

Merged

pull[bot] merged 32 commits into

Cache-Cloud:masterfrom

tensorflow:master

Sep 1, 2026

Commits on Feb 2, 2026

cmake: Exclude subdirectories from all builds
tdarote
committed
929901d

Commits on Mar 9, 2026

Document shuffle + zip interaction in tf.data.Dataset
VibhorGautam
committed
b31faa1

Commits on Jul 4, 2026

Remove fragile seed-matching example, recommend shuffle-after-zip only
VibhorGautam
committed
4ee6562

Commits on Jul 13, 2026

Merge branch 'master' into fix/shuffle-zip-docs
VibhorGautam
authored
7e7ad97

Commits on Jul 24, 2026

Fix signed-to-unsigned conversion bug in DatasetRandomAccessCache::Get
Deeven-Seru
committed
ed8d4c1

Commits on Jul 25, 2026

Fix 32-bit index truncation issues per reviewer feedback
Deeven-Seru
committed
3324146

Commits on Jul 29, 2026

[tf.data] Validate `buffer_size` in shuffle to prevent process crash
tirthfx
committed
c46fe7a

Commits on Jul 31, 2026

Address review: validate buffer_size via constant_value, not isinstance
tirthfx
committed
66bf279

Commits on Aug 23, 2026

Allow a Tensor buffer_size in Dataset.prefetch
abhijeet117
committed
7d43384

Commits on Aug 27, 2026

Reject non-empty tensors missing tensor_content in ParseFast
vee1e
committed
2bbfb12

Commits on Aug 28, 2026

Update tensor_coding test assertions for ParseSlow fallback and amplification
vee1e
committed
8aebb14

Commits on Aug 31, 2026

Merge pull request #111928 from VibhorGautam:fix/shuffle-zip-docs
tensorflower-gardener
committed
1c93abb
Use the new XLA GPU autotune cache format by default.

shyamliagrawal
authored and
tensorflower-gardener
committed
5121c32
Remove StreamExecutorGpuClient and update all references to use CommonPjRtClient instead.

pschuh
authored and
tensorflower-gardener
committed
ad3cdcf
Prevent fusing AllGather instructions into GEMM fusions.

vwbaker
authored and
tensorflower-gardener
committed
fa0bc06
Automated Code Change
tensorflower-gardener
committed
75cafaf
Allow visibility of profile_interface and profile_factory prior torch_tpu migration layout

ecalubaquib
authored and
tensorflower-gardener
committed
e059531
Fix heap buffer overflow in DynamicUpdateSliceInt4 by clamping memcpy length.
tensorflower-gardener
committed
dbff666
Merge pull request #126283 from vee1e:fix/tensor-coding-uninitialized-allocation
tensorflower-gardener
committed
3da48b3
Merge pull request #125956 from abhijeet117:fix/prefetch-tensor-buffer-size
tensorflower-gardener
committed
2c5ad30
Merge pull request #123491 from Deeven-Seru:fix-cache-dataset-negative-index
tensorflower-gardener
committed
983dd6e
Automated Code Change
tensorflower-gardener
committed
94f18a7
Add FourSigFigs for rounding step time outputs.

bmass02
authored and
tensorflower-gardener
committed
3f3bfd6
Merge pull request #124248 from tirthfx:fix/shuffle-dataset-buffer-size-validation
tensorflower-gardener
committed
7223a98
PR #47666: [XLA:GPU][oneAPI] Add XPU target check in Triton ThreadDim extraction

akhilgoe
authored and
tensorflower-gardener
committed
8a1b178
Propagate actual failure status to definition event promises in AsyncHostToDeviceTransferManager.

junwhanahn
authored and
tensorflower-gardener
committed
afc9e6a
Don't DCE tokens in opt barriers. This allows expressing an after relation like

pschuh
authored and
tensorflower-gardener
committed
a24bc35
Add BlockScalingConfig to xla_data.proto and hlo.proto to describe block
tensorflower-gardener
committed
66420a1
[Mosaic TPU] Support packed types strided load with static indices by emulation.

yueshengys
authored and
tensorflower-gardener
committed
eafe265
Merge pull request #109378 from tdarote:optimization-exclude-subdirs
tensorflower-gardener
committed
cb2bb36
Switch PJRT C API GPU client to use StreamExecutor client directly.

zacmustin
authored and
tensorflower-gardener
committed
505c29c

Commits on Sep 1, 2026

[IFRT IR] Add Fingerprint method to IfrtIRProgram

ICGog
authored and
tensorflower-gardener
committed
3e1e9b9