Commits

Breadcrumbs

History for

onroadmap-update

User selector

All users

All time

Commit History

Commits on Nov 3, 2016

• Merge commit for internal changes

  

  Vijay Vasudevan

  committed
  b96158c

  Copy full SHA for b96158c

• Seal gradients interface.

  Show description for 36621ac
  

  Patrick Nguyen

  authored and

  tensorflower-gardener

  committed
  36621ac

  Copy full SHA for 36621ac

Commits on Nov 2, 2016

• Hessian with respect to one-dimensional tensors (#5329)

  Show description for 5dd28f3
  

  tillahoffmann

  authored and

  Vijay Vasudevan

  committed
  5dd28f3

  Copy full SHA for 5dd28f3

Commits on Oct 28, 2016

• Propogate attrs from the forward function call

  Show description for f2e46bd
  tensorflower-gardener

  committed
  f2e46bd

  Copy full SHA for f2e46bd

Commits on Sep 9, 2016

• Removes a few limitation of Defun:

  Show description for 1d2aa94
  tensorflower-gardener

  committed
  1d2aa94

  Copy full SHA for 1d2aa94

Commits on Sep 2, 2016

• Add float16, complex64, complex128 to trainable data types.

  Show description for a87d665
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  a87d665

  Copy full SHA for a87d665

Commits on Aug 31, 2016

• This CL implements the following behavior for stop_gradient in while loop gradient computation. Consider a while loop with inputs xs = (x1, x2, ..., xn) and outputs ys = (y1, y2, ..., yn). If all t…

  Show description for c47a337
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  c47a337

  Copy full SHA for c47a337

Commits on Aug 9, 2016

• Deprecate op_scope by extending name_scope and move all callers to the new name_scope.

  Show description for 1df3fb0
  

  Lukasz Kaiser

  authored and

  tensorflower-gardener

  committed
  1df3fb0

  Copy full SHA for 1df3fb0

Commits on Aug 1, 2016

• Don't consider control inputs in gradient backprop graph construction. I added

  Show description for 9d7c949
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  9d7c949

  Copy full SHA for 9d7c949

Commits on Jul 24, 2016

• Adds python_grad_func argument to function.Defun.

  Show description for 01925fe
  tensorflower-gardener

  committed
  01925fe

  Copy full SHA for 01925fe

Commits on Jul 14, 2016

• Get all the gradients for a while_loop properly colocated with the ops.

  Show description for 763f8b0
  

  ebrevdo

  authored and

  tensorflower-gardener

  committed
  763f8b0

  Copy full SHA for 763f8b0

Commits on Jun 20, 2016

• Fix a bug related to using stop_gradient inside a loop.

  Show description for 42496a1
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  42496a1

  Copy full SHA for 42496a1

Commits on Jun 16, 2016

• Aggregate gradients in each device first.

  Show description for 88a50ea
  tensorflower-gardener

  committed
  88a50ea

  Copy full SHA for 88a50ea

Commits on Jun 14, 2016

• Execute TODOs to

  Show description for a558c6e
  

  itsmeolivia

  authored and

  tensorflower-gardener

  committed
  a558c6e

  Copy full SHA for a558c6e

Commits on Jun 2, 2016

• Update copyright for 3p/tf/python.

  Show description for 0cf9ed3
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  0cf9ed3

  Copy full SHA for 0cf9ed3

Commits on May 31, 2016

• Bug fix in loop gradient computation. When the value of a loop variable at iteration i+1 doesn't depend on its value at iteration i, we need to use zeros for the partial gradients of all iterations…

  Show description for 98e1214
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  98e1214

  Copy full SHA for 98e1214

Commits on Apr 22, 2016

• Rename python/platform/logging.py to python/platform/tf_logging to simplify future platform refactorings

  Show description for 455a27f
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  455a27f

  Copy full SHA for 455a27f

Commits on Apr 16, 2016

• Add shape information to dense gradients.

  Show description for f88d0cb
  

  mrry

  authored and

  tensorflower-gardener

  committed
  f88d0cb

  Copy full SHA for f88d0cb

Commits on Apr 14, 2016

• Extend singleton case in _AggregatedGrads to IndexedSlices

  Show description for a58e5ca
  

  girving

  authored and

  tensorflower-gardener

  committed
  a58e5ca

  Copy full SHA for a58e5ca

Commits on Mar 30, 2016

• Fix short description of gradients function to match actual behavior.

  Show description for 1c8a7a3
  

  yaroslavvb

  authored and

  tensorflower-gardener

  committed
  1c8a7a3

  Copy full SHA for 1c8a7a3

Commits on Mar 23, 2016

• In backprop, we may create zero tensors for the gradient of cond. Before this CL, the zero tensors are created unconditionally. This CL changes it to create the tensors only when they are needed.

  Show description for ca62bbc
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  ca62bbc

  Copy full SHA for ca62bbc

Commits on Mar 11, 2016

• Implements "bool strictness" for `tf.Tensor`.

  Show description for f6f77ac
  

  mrry

  authored and

  tensorflower-gardener

  committed
  f6f77ac

  Copy full SHA for f6f77ac

Commits on Mar 9, 2016

• Further improvements for training long sequences. Here is the current performance of rnn.dyanmic_rnn() on various sequence lengths. The last column is the time for each time step. It stayed as a co…

  Show description for 204400d
  

  yuanbyu

  authored and

  tensorflower-gardener

  committed
  204400d

  Copy full SHA for 204400d

Commits on Mar 2, 2016

• TensorFlow: add "colocate_with" feature that uses _class attribute.

  Show description for 115d41f
  

  Vijay Vasudevan

  authored and

  tensorflower-gardener

  committed
  115d41f

  Copy full SHA for 115d41f

Commits on Feb 11, 2016

• Get rid of some import cruft.

  Show description for 7760ce5
  

  josh11b

  authored and

  tensorflower-gardener

  committed
  7760ce5

  Copy full SHA for 7760ce5

Commits on Feb 6, 2016

• Consolidate the device function and device string handling in `tf.device()`.

  Show description for 3fa9676
  

  mrry

  authored and

  keveman

  committed
  3fa9676

  Copy full SHA for 3fa9676

Commits on Jan 21, 2016

• Prepare to hide tf.tensor_util

  Show description for 877fcd1
  

  girving

  authored and

  keveman

  committed
  877fcd1

  Copy full SHA for 877fcd1

Commits on Jan 20, 2016

• Add support for Refs in control flow ops (Merge, Switch, Enter, Exit).

  Show description for 7ffa2d7
  

  ebrevdo

  authored and

  keveman

  committed
  7ffa2d7

  Copy full SHA for 7ffa2d7

• Change: 112499684

  

  A. Unique TensorFlower

  authored and

  keveman

  committed
  b916d65

  Copy full SHA for b916d65

Commits on Jan 18, 2016

• Improve gradient support for nested while loops and conditionals.

  Show description for 48a95ed
  

  A. Unique TensorFlower

  authored and

  keveman

  committed
  48a95ed

  Copy full SHA for 48a95ed

Commits on Jan 16, 2016

• Change: 112277705

  

  A. Unique TensorFlower

  authored and

  keveman

  committed
  0a43371

  Copy full SHA for 0a43371

Commits on Dec 10, 2015

• TensorFlow: Upstream changes from git.

  Show description for 2725935
  

  Vijay Vasudevan

  committed
  2725935

  Copy full SHA for 2725935

Commits on Dec 7, 2015

• TensorFlow: upstrea changes from git.

  Show description for cd53f3c
  

  Vijay Vasudevan

  committed
  cd53f3c

  Copy full SHA for cd53f3c

Commits on Dec 6, 2015

• TensorFlow: upstream latest changes to git.

  Show description for f9d3e9d
  

  Vijay Vasudevan

  committed
  f9d3e9d

  Copy full SHA for f9d3e9d

Commits on Dec 1, 2015

• TensorFlow: upstream changes to git

  Show description for 795f35d
  

  Vijay Vasudevan

  committed
  795f35d

  Copy full SHA for 795f35d