Sunbelt Computer Software

PL/B Language Development and Support

Commits · Infinity-developer-test/tensorflow · GitHub

Commits

Commits on May 28, 2026

Commits on May 27, 2026

No public description.

sirakiin
authored and
tensorflower-gardener
committed
Reverts 7b1473ec2c3d49a8512308c69149886c73378e74

KanishAnand
authored and
tensorflower-gardener
committed
Fix null pointer dereference in TF_GraphImportGraphDefWithReturnOutputs on failure.

patnotz
authored and
tensorflower-gardener
committed
add constructor without enable_llvm_module_compilation_parallelism to CompilationProviderOptions

ermilovmaxim
authored and
tensorflower-gardener
committed
[XLA] Recalculate ExponentWidth based on exponent range.

majnemer
authored and
tensorflower-gardener
committed
Pass LoadOptions across PJRT C API during executable load

hhb
authored and
tensorflower-gardener
committed
Update test for ClearMemoryStats to allocate larger memory to prevent alignment issues for some backends.

bhavya01
authored and
tensorflower-gardener
committed
Remove redundant register moves in the 4-bit fully connected kernel assembly, use the pointers in the load/store instructions directly.

lrdxgm
authored and
tensorflower-gardener
committed
Improve error handling and resource cleanup in TPU profiler C API.

dmiltr3
authored and
tensorflower-gardener
committed
Add user guide for the `HLO Isolation Test` tool
tensorflower-gardener
committed
Move PjRtRegisterCompiler() into the test, since it's not used elsewhere.

dlee-dlee
authored and
tensorflower-gardener
committed
Optimize 4-bit fully connected kernel on ARM32 using 128-bit NEON registers: use q-registers (128-bit) for bitwise AND operations, replacing multiple d-register (64-bit) operations.

lrdxgm
authored and
tensorflower-gardener
committed
Fix DoS and uninitialized memory leak in LoadAndRemapMatrixOp.

patnotz
authored and
tensorflower-gardener
committed
[SDY] Update PjRt MLIR/HLO conversion APIs to accept a SDY target version during serialization

hyeontaek
authored and
tensorflower-gardener
committed
Update `ml-build` Dockerfile to support StableHLO CI workflows

mrguenther
authored and
tensorflower-gardener
committed
Use more efficient NEON instructions for initializing registers to zero.

lrdxgm
authored and
tensorflower-gardener
committed
Disable all epilogue fusions for NCHW convolutions in cuDNN fusion pipeline.

derdrdirk
authored and
tensorflower-gardener
committed
Add zizmor ignore comments for ml-build container images
tensorflower-gardener
committed
Simplify backend config to use one-of proto in autotuner.

shyamliagrawal
authored and
tensorflower-gardener
committed
Remove SpmdPartitioningAllShardingTest fixture and migrate all tests to use a fully-parameterised SpmdPartitioningTest.

Kingston Mandisodza
authored and
tensorflower-gardener
committed
Integrate LLVM at llvm/llvm-project@a225aafbd1a4
tensorflower-gardener
committed
[XLA:GPU] Initialize members of SymbolicMap to avoid UB and segfaults.

reichlfl
authored and
tensorflower-gardener
committed
Add validation for argument_index in KernelArgPackingRelocation::FromProto.

dmiltr3
authored and
tensorflower-gardener
committed
Update HDF5 plugin path handling in Keras.

dmiltr3
authored and
tensorflower-gardener
committed
[XLA:GPU] Use kCollectiveMemory instead of kP2P since at this point these memory spaces are identical.

PatriosTheGreat
authored and
tensorflower-gardener
committed