Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit History

Commits on Dec 18, 2017

• Support non-const paddings.

  Show description for 845fe8d
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  845fe8d

  Copy full SHA for 845fe8d

Commits on Dec 15, 2017

• Added a python API to estimate the performance of HW devices

  Show description for 5582df1
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  5582df1

  Copy full SHA for 5582df1

• Made the list of op kernels available.

  Show description for 81fe1b8
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  81fe1b8

  Copy full SHA for 81fe1b8

• Fix missing port number in a few places for shape checking.

  Show description for 9a6cd47
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  9a6cd47

  Copy full SHA for 9a6cd47

Commits on Dec 14, 2017

• Added the ability to sort important ops topologically

  Show description for 264e7e8
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  264e7e8

  Copy full SHA for 264e7e8

• Enable neutral element optimization.

  Show description for 3b52e03
  tensorflower-gardener

  committed
  3b52e03

  Copy full SHA for 3b52e03

Commits on Dec 13, 2017

• Generate a unique name for const used by transpose node, so that we don't use a name that is previously used already. Add dependency for const in binary op if in a loop/frame.

  Show description for 3d74c71
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  3d74c71

  Copy full SHA for 3d74c71

• Support Shape and ShapeN ops.

  Show description for ea78050
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  ea78050

  Copy full SHA for ea78050

• Simplify the naming of added transpose node. Refactor dimension check and the node creation for data format transform for better code reuse (data format transform to be used for both inputs and out…

  Show description for 5ecc3bf
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  5ecc3bf

  Copy full SHA for 5ecc3bf

Commits on Dec 12, 2017

• Added a debug mode to the model analyzer to make it easier to figure out why shapes are missing.

  Show description for 389ffa8
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  389ffa8

  Copy full SHA for 389ffa8

• Disable neutral element and reciprocal optimizations again.

  Show description for 0e1c557
  tensorflower-gardener

  committed
  0e1c557

  Copy full SHA for 0e1c557

Commits on Dec 11, 2017

• Enable optimizations of operations with neutral/absorbing elements by default. We leave removal of addition and subtraction with zero out for now, since it is used as a "hack" to force a copy of a …

  Show description for a4b68b6
  tensorflower-gardener

  committed
  a4b68b6

  Copy full SHA for a4b68b6

• Always inline functions when creating an item.

  Show description for c290c47
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  c290c47

  Copy full SHA for c290c47

• Use DataFormatVecPermute instead Gather, which is very slow.

  Show description for 0117c7a
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  0117c7a

  Copy full SHA for 0117c7a

• Make sure we serialize the creation and deletion of clusters from python to avoid race conditions.

  Show description for f783790
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  f783790

  Copy full SHA for f783790

Commits on Dec 9, 2017

• Support non-const input sizes for Conv2DBackpropInput.

  Show description for 4a19c34
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  4a19c34

  Copy full SHA for 4a19c34

Commits on Dec 6, 2017

• Support non-const axis for split and concat.

  Show description for 707d586
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  707d586

  Copy full SHA for 707d586

Commits on Dec 4, 2017

• Enable dependency optimizer in Grappler by default.

  Show description for 50efd92
  tensorflower-gardener

  committed
  50efd92

  Copy full SHA for 50efd92

Commits on Nov 29, 2017

• Simplify const node creation.

  Show description for 48347ee
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  48347ee

  Copy full SHA for 48347ee

Commits on Nov 28, 2017

• Added an option to assume that the shape of fed nodes in unknown since any shape can be actually used.

  Show description for a86f2d2
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  a86f2d2

  Copy full SHA for a86f2d2

Commits on Nov 27, 2017

• Improved the item_tst.py regression test.

  Show description for f327ae0
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  f327ae0

  Copy full SHA for f327ae0

• Add the missing GPU configs to avoid the division by zero (gflops) error.

  Show description for 75350d3
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  75350d3

  Copy full SHA for 75350d3

• Delete trailing whitespace

  Show description for 191825e
  tensorflower-gardener

  committed
  191825e

  Copy full SHA for 191825e

Commits on Nov 23, 2017

• Prevented a couple of memory leaks in the code generated by swig

  Show description for 4e5534d
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  4e5534d

  Copy full SHA for 4e5534d

• Acquire the GIL before working with PyLists and PyDict.

  Show description for 8067548
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  8067548

  Copy full SHA for 8067548

Commits on Nov 22, 2017

• Turn off layout optimizer (as with all other optimizers) in the memory

  Show description for d390d27
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  d390d27

  Copy full SHA for d390d27

• Let python handle the deletion of the item objects instead of trying to take

  Show description for e6840e8
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  e6840e8

  Copy full SHA for e6840e8

• Support depthwise conv ops.

  Show description for 6ad9e45
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  6ad9e45

  Copy full SHA for 6ad9e45

Commits on Nov 21, 2017

• Pass VirtualCluster to tf_optimizer to enable cross-optimization of models.

  Show description for f0d1abb
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  f0d1abb

  Copy full SHA for f0d1abb

• Update the tf_item whenever the underlying metagraph is updated

  Show description for a784520
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  a784520

  Copy full SHA for a784520

• Added the ability to report peak memory usage

  Show description for bb96a30
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  bb96a30

  Copy full SHA for bb96a30

Commits on Nov 20, 2017

• Added an option to make the graph optimizer less verbose

  Show description for 7479dc0
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  7479dc0

  Copy full SHA for 7479dc0

Commits on Nov 18, 2017

• Added the ability to list the hardware resources available in a cluster.

  Show description for 63a69b8
  

  benoitsteiner

  authored and

  tensorflower-gardener

  committed
  63a69b8

  Copy full SHA for 63a69b8

Commits on Nov 15, 2017

• Update layout_optimizer gradient test to use a variable for the input instead of a constant. Otherwise the dependency optimizer eliminates the Conv2DBackpropInput.

  Show description for 0c9b03c
  tensorflower-gardener

  committed
  0c9b03c

  Copy full SHA for 0c9b03c

Commits on Nov 14, 2017

• Use Toggle instead of bool to make the layout optimizer name and usage consistent with other optimizers.

  Show description for 6357baf
  

  Yao Zhang

  authored and

  tensorflower-gardener

  committed
  6357baf

  Copy full SHA for 6357baf