All pull requests

Pull requests

• [TF] Switch TensorFlow default build to Bzlmod and sync MODULE.bazel.

  #129936·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Lower SRQ tfl.batch_matmul with blockwise weights to blockwise RHS.

  #129935·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  

• Validate number attrs and empty resource inputs in function.cc

  size:M
  #129934·professor-moody opened Nov 3, 2026Contributor·

  ·
  2

• PR #52165: [XLA-CPU][oneDNN] Update oneDNN to v3.14.1

  #129933·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  1

• Migrate tensorflow/compiler/jit execution and XlaDevice unconditionally to PJRT

  #129930·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Unify core/common_runtime/gpu and eager on PJRT, optimize GPUUtil PjRtTensorBuffer transfers, and forward CPU intra-op thread pool

  #129929·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Integrate LLVM at llvm/llvm-project@fd4b4d45e3d8

  #129928·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Only statically unpack affine quantized int8 DEQUANTIZE inputs in XNNPack.

  #129926·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  

• fix(math_ops): fix dtype mismatch in tf.pow with mixed integer and float arguments

  awaiting reviewsize:S
  #129924·vivekjetani opened Nov 3, 2026·

  ·
  1

  14

• PR #52165: [XLA-CPU][oneDNN] Update oneDNN to v3.14.1

  #129923·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  1

• Support SparseTensorValue in tf.sparse.bincount and validate_sparse_weights

  awaiting reviewsize:M
  #129922·Hrushikesh-ramilla opened Nov 3, 2026Contributor·

  ·
  1

  4

• fix(kernels): support CallFrameInterface and host memory attributes in ForOp

  size:M
  #129920·genius-0963 opened Nov 3, 2026Contributor·

  ·
  1

  5

• PR #51226: [ROCm] Bump up to theRock10.1.0

  #129919·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  1

• Reject non-affine quantized inputs in the builtin BATCH_MATMUL kernel.

  #129918·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  

• [Sparse] Validate input and convert SparseTensorValue in tf.sparse.reduce_sum and tf.sparse.reduce_max

  awaiting reviewsize:M
  #129917·Hrushikesh-ramilla opened Nov 3, 2026Contributor·

  ·
  1

  3

• Avoid temporary string allocations in NodeDef attribute lookup and insertion.

  #129915·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Speed up TensorShapeBase::AsProto and TensorShapeBase::dim_sizes for all representations.

  #129914·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Automated Code Change

  #129913·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Reject unimplemented quantizations in the builtin DEQUANTIZE kernel.

  #129912·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  

• Apply the fused activation of blockwise quantized FULLY_CONNECTED ops.

  #129911·copybara-service[bot] opened Nov 3, 2026Bot·

  ·
  

• Automated Code Change

  #129910·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Support SparseTensorValue and validate input in tf.sparse.softmax

  awaiting reviewkokoro:force-runprtype:bugfixpythonready to pullsize:S
  #129909·Hrushikesh-ramilla opened Nov 3, 2026Contributor·

  ·
  1

  

• Automated Code Change

  #129907·copybara-service[bot] opened Nov 3, 2026Bot·

  ·

• Support SparseTensorValue and validate input in tf.sparse.maximum and tf.sparse.minimum

  awaiting reviewkokoro:force-runprtype:bugfixpythonready to pullsize:S
  #129906·Hrushikesh-ramilla opened Nov 3, 2026Contributor·

  ·
  1

  4

  

• validate convolution dimension_numbers form a permutation

  awaiting reviewcomp:coresize:M
  #129903·aysha-afrah26 opened Nov 3, 2026Contributor·

  ·
  2