Commits

Breadcrumbs

History for

onroadmap-update

User selector

All users

All time

Commit History

Commits on Jan 4, 2017

• Adds V2 versions of Queue and Reader ops using ResourceHandles.

  Show description for 2482564
  tensorflower-gardener

  committed
  2482564

  Copy full SHA for 2482564

Commits on Nov 28, 2016

• Uses containers for per_step resources instead of a separate resource_mgr.

  Show description for b571f2e
  tensorflower-gardener

  committed
  b571f2e

  Copy full SHA for b571f2e

Commits on Nov 9, 2016

• Merge changes from github.

  Show description for a771598
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  a771598

  Copy full SHA for a771598

Commits on Nov 3, 2016

• Allow Android to use legacy scalars for parity in model compatibility.

  Show description for 9974262
  

  andrewharp

  authored and

  tensorflower-gardener

  committed
  9974262

  Copy full SHA for 9974262

Commits on Oct 20, 2016

• Merge changes from github.

  Show description for c5ab3dd
  

  Patrick Nguyen

  authored and

  tensorflower-gardener

  committed
  c5ab3dd

  Copy full SHA for c5ab3dd

Commits on Oct 8, 2016

• Automated rollback of change 135516295

  Show description for 3510b30
  tensorflower-gardener

  committed
  3510b30

  Copy full SHA for 3510b30

Commits on Oct 7, 2016

• Change selective registration to avoid use of strstr and strcmp that are not

  Show description for c37847c
  tensorflower-gardener

  committed
  c37847c

  Copy full SHA for c37847c

Commits on Sep 20, 2016

• Remove some unnecessary todos and move some unrecommended functions

  Show description for 11230eb
  

  Vijay Vasudevan

  authored and

  tensorflower-gardener

  committed
  11230eb

  Copy full SHA for 11230eb

Commits on Aug 10, 2016

• In kernel registration when selective_registration is enabled, also

  Show description for 331a2fa
  tensorflower-gardener

  committed
  331a2fa

  Copy full SHA for 331a2fa

Commits on Aug 5, 2016

• More verbose errors for unregistered kernels.

  Show description for d2b5fb9
  tensorflower-gardener

  committed
  d2b5fb9

  Copy full SHA for d2b5fb9

Commits on Jul 29, 2016

• Make sure that kernels that are meant to run on GPU are scheduled inline. This

  Show description for f1acb3b
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  f1acb3b

  Copy full SHA for f1acb3b

Commits on Jul 18, 2016

• Reduce overhead of running kernels:

  Show description for b8ccc5a
  tensorflower-gardener

  committed
  b8ccc5a

  Copy full SHA for b8ccc5a

Commits on Jun 30, 2016

• Fix initialization issues with Variables whose shape contains a zero.

  Show description for 379df09
  tensorflower-gardener

  committed
  379df09

  Copy full SHA for 379df09

• Merge changes from github.

  Show description for 533d891
  tensorflower-gardener

  committed
  533d891

  Copy full SHA for 533d891

Commits on Jun 2, 2016

• Update copyright for 3p/tf/core.

  Show description for c8b59c0
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  c8b59c0

  Copy full SHA for c8b59c0

Commits on Jun 1, 2016

• Change function library to use a runner passed in from the Run call, instead of

  Show description for caed96e
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  caed96e

  Copy full SHA for caed96e

Commits on May 17, 2016

• Avoid calling LogMemory::IsEnabled in allocate_tensor. This saves a couple

  Show description for 20c7eca
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  20c7eca

  Copy full SHA for 20c7eca

Commits on May 4, 2016

• Only construct UniqueTensorReferences if they are needed.

  Show description for fc99eea
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  fc99eea

  Copy full SHA for fc99eea

Commits on Apr 21, 2016

• Check for op deprecation in C++ API and graph validation

  Show description for 18429eb
  

  girving

  authored and

  tensorflower-gardener

  committed
  18429eb

  Copy full SHA for 18429eb

Commits on Apr 20, 2016

• Add a .Deprecated method to REGISTER_OP

  Show description for 30334d2
  

  girving

  authored and

  tensorflower-gardener

  committed
  30334d2

  Copy full SHA for 30334d2

Commits on Apr 11, 2016

• Make the TF_CALL_*() macros drop the semicolon, instead putting that burden

  Show description for 6e8e7ca
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  6e8e7ca

  Copy full SHA for 6e8e7ca

Commits on Apr 10, 2016

• This is another step to make TensorFlow more interactive and flexible to users. It allows a tensor produced by a run call to stay "in-place" so that a future run call can use it in-place. To achiev…

  Show description for 098f930
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  098f930

  Copy full SHA for 098f930

Commits on Mar 28, 2016

• Change selective registration of op kernels to work off of the class name

  Show description for e30d80d
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  e30d80d

  Copy full SHA for e30d80d

Commits on Mar 25, 2016

• Mention OpKernel thread-safety requirements.

  Show description for c9b94f0
  

  rryan

  authored and

  tensorflower-gardener

  committed
  c9b94f0

  Copy full SHA for c9b94f0

Commits on Mar 24, 2016

• Several changes to reduce TensorFlow code size (especially important

  Show description for 821920c
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  821920c

  Copy full SHA for 821920c

Commits on Mar 22, 2016

• Apply selective registration to op registration, to remove the OpDefBuilder

  Show description for 03993ff
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  03993ff

  Copy full SHA for 03993ff

Commits on Mar 12, 2016

• Add optional cc file whitelisting for operator kernel registration.

  Show description for eb56025
  

  andrewharp

  authored and

  tensorflower-gardener

  committed
  eb56025

  Copy full SHA for eb56025

Commits on Mar 9, 2016

• Assume errors are rare in OP_REQUIRES*() macros.

  Show description for 01b60af
  

  josh11b

  authored and

  tensorflower-gardener

  committed
  01b60af

  Copy full SHA for 01b60af

Commits on Mar 2, 2016

• Changes to reduce code size (especially useful for mobile):

  Show description for b76812c
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  b76812c

  Copy full SHA for b76812c

Commits on Feb 26, 2016

• Add a step_id field to the OpKernelContext indicating which step

  Show description for e8260f3
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  e8260f3

  Copy full SHA for e8260f3

Commits on Feb 25, 2016

• Avoid some over-inlined routines. Reduces code size of TensorFlow binaries

  Show description for 9ccc4b6
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  9ccc4b6

  Copy full SHA for 9ccc4b6

Commits on Feb 22, 2016

• Added the ability to track allocation sizes in the tracking allocator itself if the underlying

  Show description for 8f8f7f1
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  8f8f7f1

  Copy full SHA for 8f8f7f1

Commits on Feb 17, 2016

• Merge changes from github.

  Show description for fe056f0
  

  Vijay Vasudevan

  authored and

  tensorflower-gardener

  committed
  fe056f0

  Copy full SHA for fe056f0

Commits on Feb 16, 2016

• Several changes to reduce the overhead and the cache footprint of dispatching

  Show description for cf66101
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  cf66101

  Copy full SHA for cf66101

Commits on Feb 11, 2016

• Moves MemoryType inference code out of OpKernel so that it can reused.

  Show description for 7415dfa
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  7415dfa

  Copy full SHA for 7415dfa