Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit History

Commits on Jan 11, 2017

• More numerically stable gradients for division. (#6562)

  

  tillahoffmann

  authored and

  drpngx

  committed
  d4b5c60

  Copy full SHA for d4b5c60

• Switch tf.concat_v2 references in third_party/tensorflow to tf.concat.

  Show description for 0e226af
  tensorflower-gardener

  committed
  0e226af

  Copy full SHA for 0e226af

Commits on Dec 30, 2016

• Add missing gradients to tf.ceil and tf.round.

  Show description for 7455e25
  tensorflower-gardener

  committed
  7455e25

  Copy full SHA for 7455e25

Commits on Dec 20, 2016

• Rename usages of tf.mul, tf.neg, tf.sub that are used internally

  Show description for cb4acf5
  

  aselle

  authored and

  tensorflower-gardener

  committed
  cb4acf5

  Copy full SHA for cb4acf5

Commits on Dec 15, 2016

• Add Expm1 Op.

  Show description for 6731e21
  tensorflower-gardener

  committed
  6731e21

  Copy full SHA for 6731e21

Commits on Dec 7, 2016

• Remove remaining uses of tf.reverse() and remove tf.reverse()

  Show description for 878f5fc
  

  aselle

  authored and

  tensorflower-gardener

  committed
  878f5fc

  Copy full SHA for 878f5fc

• Switch all tf.concat(concat_dim, value, name) calls in third_party/tensorflow to tf.concat_v2(value, axis, name).

  Show description for d4eb834
  tensorflower-gardener

  committed
  d4eb834

  Copy full SHA for d4eb834

Commits on Nov 30, 2016

• Deprecate tf.select since it is getting replaced by tf.where.

  Show description for fe558b0
  tensorflower-gardener

  committed
  fe558b0

  Copy full SHA for fe558b0

Commits on Nov 22, 2016

• Change division in TensorFlow to flooring semantics.

  Show description for fcc3923
  

  aselle

  authored and

  tensorflower-gardener

  committed
  fcc3923

  Copy full SHA for fcc3923

Commits on Nov 21, 2016

• Deprecate tf.batch_matmul and replace with equivalent calls to tf.matmul that now supports adjoint and batch matmul.

  Show description for 73d11f4
  tensorflower-gardener

  committed
  73d11f4

  Copy full SHA for 73d11f4

Commits on Nov 17, 2016

• Merge changes from github.

  Show description for 54e5000
  tensorflower-gardener

  committed
  54e5000

  Copy full SHA for 54e5000

Commits on Nov 15, 2016

• Deprecate tf.inv in favor of tf.reciprocal.

  Show description for fb01ebb
  

  martinwicke

  authored and

  tensorflower-gardener

  committed
  fb01ebb

  Copy full SHA for fb01ebb

Commits on Nov 9, 2016

• Remove uses and deprecate listdiff (use setdiff1d)

  Show description for 1d0bbea
  

  aselle

  authored and

  tensorflower-gardener

  committed
  1d0bbea

  Copy full SHA for 1d0bbea

Commits on Nov 4, 2016

• Merge changes from github.

  Show description for 818993c
  

  Vijay Vasudevan

  authored and

  tensorflower-gardener

  committed
  818993c

  Copy full SHA for 818993c

Commits on Sep 15, 2016

• Consolidate fft ops: Replace fft* ops with their batch_fft* equivalents and deprecate the latter.

  Show description for 9287c2f
  tensorflower-gardener

  committed
  9287c2f

  Copy full SHA for 9287c2f

Commits on Sep 10, 2016

• Merge changes from github.

  Show description for 54a7178
  

  Dan Mané

  authored and

  tensorflower-gardener

  committed
  54a7178

  Copy full SHA for 54a7178

Commits on Sep 9, 2016

• Rename NoGradient -> NotDifferentiable, to make it clear

  Show description for 612bae7
  

  Vijay Vasudevan

  authored and

  tensorflower-gardener

  committed
  612bae7

  Copy full SHA for 612bae7

Commits on Sep 7, 2016

• Add gradient for RsqrtGrad. Change test to use mixed relative + absolute tolerance, which allows tightening tolerances a bit.

  Show description for 970c408
  tensorflower-gardener

  committed
  970c408

  Copy full SHA for 970c408

Commits on Sep 4, 2016

• Add gradient for InvGrad.

  Show description for b33f0b7
  tensorflower-gardener

  committed
  b33f0b7

  Copy full SHA for b33f0b7

Commits on Sep 3, 2016

• Add gradients for TanhGrad and SigmoidGrad.

  Show description for 24b302d
  tensorflower-gardener

  committed
  24b302d

  Copy full SHA for 24b302d

Commits on Sep 2, 2016

• Remove NoGradient for ops which could have gradients defined but don't.

  Show description for 6c110d3
  

  Mrinal Kalakrishnan

  authored and

  tensorflower-gardener

  committed
  6c110d3

  Copy full SHA for 6c110d3

• Fix gradients of numerous complex ops

  Show description for 821063d
  

  girving

  authored and

  tensorflower-gardener

  committed
  821063d

  Copy full SHA for 821063d

• Put CPU-only ops in ProdGrad in a device('/cpu:0') block

  Show description for 182fef1
  

  girving

  authored and

  tensorflower-gardener

  committed
  182fef1

  Copy full SHA for 182fef1

Commits on Aug 31, 2016

• Define a gradient for the SqrtGrad op.

  Show description for b37153c
  tensorflower-gardener

  committed
  b37153c

  Copy full SHA for b37153c

Commits on Aug 27, 2016

• Optimized the gradients of the sqrt, rsqrt, and inv functions

  Show description for f2f582b
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  f2f582b

  Copy full SHA for f2f582b

Commits on Aug 25, 2016

• Fix gradient of pow for complex types

  Show description for e78cbe0
  tensorflower-gardener

  committed
  e78cbe0

  Copy full SHA for e78cbe0

Commits on Aug 12, 2016

• Merge changes from github.

  Show description for a92da9e
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  a92da9e

  Copy full SHA for a92da9e

Commits on Jul 29, 2016

• Use a division directly instead of multiplying by the inverse. Also use the rsqrt operation instead of taking the inverse of the square root.

  Show description for aebdf77
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  aebdf77

  Copy full SHA for aebdf77

Commits on Jul 25, 2016

• Merge changes from github.

  Show description for 21716d8
  

  martinwicke

  authored and

  tensorflower-gardener

  committed
  21716d8

  Copy full SHA for 21716d8

Commits on Jul 18, 2016

• Fix tf.pow gradient at zero

  Show description for a33ebe0
  

  girving

  authored and

  tensorflower-gardener

  committed
  a33ebe0

  Copy full SHA for a33ebe0

Commits on Jul 7, 2016

• Add missing ops.NoGradient() calls for new TanhGrad and SigmoidGrad ops since these ops do not have gradients themselves.

  Show description for 92d70ee
  tensorflower-gardener

  committed
  92d70ee

  Copy full SHA for 92d70ee

Commits on Jun 30, 2016

• Improved the gradients for tanh and sigmoid. This improves the speed of the ptb word model from 6800 to 7800 words per second.

  Show description for 5a8faef
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  5a8faef

  Copy full SHA for 5a8faef

Commits on Jun 25, 2016

• Merge changes from github.

  Show description for 67324b1
  tensorflower-gardener

  committed
  67324b1

  Copy full SHA for 67324b1

Commits on Jun 16, 2016

• Fast path for reduction ops when reducing all dimensions.

  Show description for d57d349
  

  Zongheng Yang

  authored and

  tensorflower-gardener

  committed
  d57d349

  Copy full SHA for d57d349

Commits on Jun 14, 2016

• Execute TODOs to

  Show description for a558c6e
  

  itsmeolivia

  authored and

  tensorflower-gardener

  committed
  a558c6e

  Copy full SHA for a558c6e