Commits

Branch selector

 master

User selector

ezhulenev

Datepicker

All time

Commit History

Commits on Apr 27, 2019

• [Grappler] Inline direct function calls using common_runtime implementation

  Show description for 0d14c9d
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  0d14c9d

  Copy full SHA for 0d14c9d

Commits on Apr 26, 2019

• Update Eigen to commit fb5130011791.

  Show description for b53fd76
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  b53fd76

  Copy full SHA for b53fd76

Commits on Apr 25, 2019

• Re-enable Resnet50 test is ?san.

  Show description for 83d09b2
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  83d09b2

  Copy full SHA for 83d09b2

Commits on Apr 24, 2019

• Update Eigen to commit c97a66af8736.

  Show description for e1da9e0
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  e1da9e0

  Copy full SHA for e1da9e0

• Update Eigen to commit 973d172a72e8.

  Show description for b15ac9f
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  b15ac9f

  Copy full SHA for b15ac9f

Commits on Apr 19, 2019

• Merge non-build changes from tensorflow#26722

  Show description for 6f4a1bf
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  6f4a1bf

  Copy full SHA for 6f4a1bf

• Compute BiasActivation using Eigen output kernel mechanism.

  Show description for 1dff4ff
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  1dff4ff

  Copy full SHA for 1dff4ff

Commits on Apr 17, 2019

• Add an option to keep inlined function call node targetable.

  Show description for 856139c
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  856139c

  Copy full SHA for 856139c

Commits on Apr 16, 2019

• Less strict error bound in conv_ops_test

  Show description for 9f3d53d
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  9f3d53d

  Copy full SHA for 9f3d53d

• QInt8 implementation of FusedConv2DBiasActivation on CPU.

  Show description for 4407064
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  4407064

  Copy full SHA for 4407064

Commits on Apr 15, 2019

• Add custom contraction kernel for QInt8xQInt8->QInt32 contraction.

  Show description for dfcbe0a
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  dfcbe0a

  Copy full SHA for dfcbe0a

Commits on Apr 12, 2019

• Cleanup eigen_contraction_kernel includes

  Show description for 4fc3d56
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  4fc3d56

  Copy full SHA for 4fc3d56

Commits on Apr 11, 2019

• [Grappler] Add support for FusedMatMul to remapper optimizer

  Show description for 076f6e4
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  076f6e4

  Copy full SHA for 076f6e4

• [Grappler] Add IsAnyMatMul and restrict IsMatMul to just 'MatMul' op type

  Show description for ceeea9c
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  ceeea9c

  Copy full SHA for ceeea9c

• Do not remap to FusedConv2D if activation is not supported on GPU

  Show description for 59e0788
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  59e0788

  Copy full SHA for 59e0788

• Automated rollback of commit c477100

  Show description for c38699c
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  c38699c

  Copy full SHA for c38699c

Commits on Apr 10, 2019

• [Grappler] Add support for FusedMatMul to remapper optimizer

  Show description for c477100
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  c477100

  Copy full SHA for c477100

• [Grappler] Fuse Relu6 and Elu into Conv2D

  Show description for 2ace7a7
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  2ace7a7

  Copy full SHA for 2ace7a7

Commits on Apr 9, 2019

• Test transpose_a/transpose_b combinations in FusedMatMul test.

  Show description for de141dc
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  de141dc

  Copy full SHA for de141dc

• Add activations support to FusedMatMul with BiasAdd: Relu, Relu6, Elu

  Show description for 5f3acb7
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  5f3acb7

  Copy full SHA for 5f3acb7

• Initial FusedMatMul kernel implementation: MatMul+BiasAdd.

  Show description for 2c12f4c
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  2c12f4c

  Copy full SHA for 2c12f4c

Commits on Apr 5, 2019

• [Grappler] Do not copy device set when optimizing function library

  Show description for 51a3acd
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  51a3acd

  Copy full SHA for 51a3acd

Commits on Apr 4, 2019

• Add missing break to the switch statement.

  Show description for 3f8281e
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  3f8281e

  Copy full SHA for 3f8281e

• Add support for Relu6 and Elu activations to FusedConv2D op

  Show description for c204b80
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  c204b80

  Copy full SHA for c204b80

Commits on Apr 3, 2019

• Update Eigen to commit 54b70baafc92.

  Show description for 0e8cbd4
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  0e8cbd4

  Copy full SHA for 0e8cbd4

Commits on Apr 2, 2019

• Lookup function node names for control outputs.

  Show description for 00f4109
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  00f4109

  Copy full SHA for 00f4109

• Add an option to keep lowered If/While nodes fetchable.

  Show description for 3ec468c
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  3ec468c

  Copy full SHA for 3ec468c

• Rewrite AccumulateNV2 into AddN

  Show description for 22413d6
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  22413d6

  Copy full SHA for 22413d6

Commits on Mar 29, 2019

• Reuse common_runtime function instantiation in Grappler function optimizer.

  Show description for 14a65fb
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  14a65fb

  Copy full SHA for 14a65fb

• Automated rollback of commit 92f736f

  Show description for 9e467f4
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  9e467f4

  Copy full SHA for 9e467f4

Commits on Mar 28, 2019

• Reuse common_runtime function instantiation in Grappler function optimizer.

  Show description for 92f736f
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  92f736f

  Copy full SHA for 92f736f

Commits on Mar 26, 2019

• Add a LoopExecuted NoOp during While lowering to split data and control edges.

  Show description for de69f30
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  de69f30

  Copy full SHA for de69f30

Commits on Mar 25, 2019

• Do not skip small graphs when doing function inlining

  Show description for 6aecc69
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  6aecc69

  Copy full SHA for 6aecc69

• Disable multi-device function inlining after partitioning (GraphOptimizer)

  Show description for c3865e7
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  c3865e7

  Copy full SHA for c3865e7

Commits on Mar 22, 2019

• Remove duplicated control inputs before adding edges to the graph

  Show description for bbe70e6
  

  ezhulenev

  authored and

  tensorflower-gardener

  committed
  bbe70e6

  Copy full SHA for bbe70e6

Pagination

Previous

Next