Commits

Branch selector

 master

User selector

bloops

Datepicker

All time

Commit History

Commits on Oct 19, 2019

• Add INT32 as a supported type for Einsum.

  Show description for 8738d94
  

  bloops

  authored and

  tensorflower-gardener

  committed
  8738d94

  Copy full SHA for 8738d94

Commits on Oct 3, 2019

• Temporarily remove functools32 import as it is causing some benchmark tests to break.

  Show description for cc38e56
  

  bloops

  authored and

  tensorflower-gardener

  committed
  cc38e56

  Copy full SHA for cc38e56

• Fix Mac OSX build linker error.

  Show description for b9d9383
  

  bloops

  authored and

  tensorflower-gardener

  committed
  b9d9383

  Copy full SHA for b9d9383

Commits on Oct 2, 2019

• Increase shard count for sparse_csr_matrix_ops_test.

  Show description for 7a89950
  

  bloops

  authored and

  tensorflower-gardener

  committed
  7a89950

  Copy full SHA for 7a89950

• Allow tf2xla Einsum Op to accept unary inputs.

  Show description for 1b92022
  

  bloops

  authored and

  tensorflower-gardener

  committed
  1b92022

  Copy full SHA for 1b92022

• Internal change

  Show description for 193f1d6
  

  bloops

  authored and

  tensorflower-gardener

  committed
  193f1d6

  Copy full SHA for 193f1d6

Commits on Sep 28, 2019

• Extend functionality of tf.einsum to the full NumPy spec. (part 2 of 2).

  Show description for 4eba4d5
  

  bloops

  authored and

  tensorflower-gardener

  committed
  4eba4d5

  Copy full SHA for 4eba4d5

Commits on Sep 26, 2019

• Add <numeric> standard library to fix windows build.

  Show description for ce753d7
  

  bloops

  authored and

  tensorflower-gardener

  committed
  ce753d7

  Copy full SHA for ce753d7

Commits on Sep 25, 2019

• Implement Sparse Transpose CPU kernel.

  Show description for 2e58c8b
  

  bloops

  authored and

  tensorflower-gardener

  committed
  2e58c8b

  Copy full SHA for 2e58c8b

• Allow XlaEinsum Shape Inference to handle ellipses.

  Show description for 122f369
  

  bloops

  authored and

  tensorflower-gardener

  committed
  122f369

  Copy full SHA for 122f369

• Internal change

  Show description for 8ebb126
  

  bloops

  authored and

  tensorflower-gardener

  committed
  8ebb126

  Copy full SHA for 8ebb126

Commits on Sep 21, 2019

• Add Python gradient implementation for EinsumOp.

  Show description for cf52c3d
  

  bloops

  authored and

  tensorflower-gardener

  committed
  cf52c3d

  Copy full SHA for cf52c3d

Commits on Sep 19, 2019

• Remove comment referencing deprecated parameter 'global_step'. Instead use `iterations`.

  Show description for cedd181
  

  bloops

  authored and

  tensorflower-gardener

  committed
  cedd181

  Copy full SHA for cedd181

Commits on Sep 10, 2019

• Add support for zero-sized inputs to EinsumOp.

  Show description for 82bf3e6
  

  bloops

  authored and

  tensorflower-gardener

  committed
  82bf3e6

  Copy full SHA for 82bf3e6

Commits on Sep 9, 2019

• Add TF2XLA registration for Einsum op, so that it uses XlaEinsum implementation. (Try #2)

  Show description for 55bb36f
  

  bloops

  authored and

  tensorflower-gardener

  committed
  55bb36f

  Copy full SHA for 55bb36f

• Fix a bug in gradient checker v2 for empty gradients where it led to a shape mismatch error.

  Show description for f0fd335
  

  bloops

  authored and

  tensorflower-gardener

  committed
  f0fd335

  Copy full SHA for f0fd335

Commits on Sep 5, 2019

• Automated rollback of commit fe3d481de8e10046a40c74af38d347ea74d9e111

  Show description for e4f4579
  

  bloops

  authored and

  tensorflower-gardener

  committed
  e4f4579

  Copy full SHA for e4f4579

Commits on Sep 3, 2019

• Add more dtypes to Einsum Op. Second try, as the first one broke global TAP because of long compilation times.

  Show description for 6462b0e
  

  bloops

  authored and

  tensorflower-gardener

  committed
  6462b0e

  Copy full SHA for 6462b0e

Commits on Aug 22, 2019

• Add more dtypes to EinsumOp. We've added exactly the registrations that are present in MatMul Op. In preparation for tf.einsum switch, these are all needed for existing tests to pass.

  Show description for 7a22b81
  

  bloops

  authored and

  tensorflower-gardener

  committed
  7a22b81

  Copy full SHA for 7a22b81

• Introduce scheduling strategy for ParellelForWithWorkerId which schedules work items with a fixed shard size.

  Show description for 11d9698
  

  bloops

  authored and

  tensorflower-gardener

  committed
  11d9698

  Copy full SHA for 11d9698

Commits on Jul 13, 2019

• Automated rollback of commit 560d21b68805ed3247f897b9ac2e1f203a168f8d

  Show description for 4a3cf05
  

  bloops

  authored and

  tensorflower-gardener

  committed
  4a3cf05

  Copy full SHA for 4a3cf05

Commits on Jul 12, 2019

• Add TF2XLA registration for Einsum op, so that it uses XlaEinsum implementation.

  Show description for 560d21b
  

  bloops

  authored and

  tensorflower-gardener

  committed
  560d21b

  Copy full SHA for 560d21b

Commits on Jul 10, 2019

• Add functools32 to TensorFlow dependencies.

  Show description for 82c46d7
  

  bloops

  authored and

  tensorflower-gardener

  committed
  82c46d7

  Copy full SHA for 82c46d7

Commits on Jun 27, 2019

• Add opt_einsum to TensorFlow dependencies.

  Show description for 8b01b9c
  

  bloops

  authored and

  tensorflower-gardener

  committed
  8b01b9c

  Copy full SHA for 8b01b9c

Commits on Jun 18, 2019

• Register GPU kernels for EinsumOp.

  Show description for 222df68
  

  bloops

  authored and

  tensorflower-gardener

  committed
  222df68

  Copy full SHA for 222df68

Commits on Jun 14, 2019

• Automated rollback of commit 8e8f040cec40b7fb86235702ae8b610de712dcf5

  Show description for c6177ea
  

  bloops

  authored and

  tensorflower-gardener

  committed
  c6177ea

  Copy full SHA for c6177ea

Commits on May 20, 2019

• Automated rollback of commit 8e8f040cec40b7fb86235702ae8b610de712dcf5

  Show description for eb3b645
  

  bloops

  authored and

  tensorflower-gardener

  committed
  eb3b645

  Copy full SHA for eb3b645

Commits on May 18, 2019

• Extend functionality of tf.einsum to the full NumPy spec. (part 1 of 2).

  Show description for 8e8f040
  

  bloops

  authored and

  tensorflower-gardener

  committed
  8e8f040

  Copy full SHA for 8e8f040

Commits on Apr 26, 2019

• Update Stable Name for tf.matmul.

  Show description for c65e9de
  

  bloops

  authored and

  tensorflower-gardener

  committed
  c65e9de

  Copy full SHA for c65e9de

• Add to ExpectedEnginesToBuild to reflect changes in tf.matmul creating another

  Show description for 580e0f6
  

  bloops

  authored and

  tensorflower-gardener

  committed
  580e0f6

  Copy full SHA for 580e0f6

Commits on Apr 25, 2019

• Add support for full BatchMatMulV2 functionality (broadcasting) in TFLite.

  Show description for 17db87c
  

  bloops

  authored and

  tensorflower-gardener

  committed
  17db87c

  Copy full SHA for 17db87c

Commits on Apr 24, 2019

• Move MatMulBCast class to core/util.

  Show description for b199a97
  

  bloops

  authored and

  tensorflower-gardener

  committed
  b199a97

  Copy full SHA for b199a97

Commits on Apr 19, 2019

• Push back compatibility window of BatchMatMulV2 Op by one more week, as more time may be needed to fix related breakages.

  Show description for 1e2c3fe
  

  bloops

  authored and

  tensorflower-gardener

  committed
  1e2c3fe

  Copy full SHA for 1e2c3fe

Commits on Apr 18, 2019

• Add C++ Gradient for BatchMatMulV2.

  Show description for 7bdabf5
  

  bloops

  authored and

  tensorflower-gardener

  committed
  7bdabf5

  Copy full SHA for 7bdabf5

Commits on Mar 29, 2019

• Add broadcasting support to tf.matmul.

  Show description for 47ab68d
  

  bloops

  authored and

  tensorflower-gardener

  committed
  47ab68d

  Copy full SHA for 47ab68d

Pagination

Previous

Next