Commits

Breadcrumbs

History for

onv

User selector

All users

All time

Commit History

Commits on Jan 5, 2017

• Change arg order for {softmax,sparse_softmax,sigmoid}_cross_entropy_with_logits to be (labels, predictions), and force use of named args to avoid accidents.

  Show description for 333dc32
  

  martinwicke

  authored and

  tensorflower-gardener

  committed
  333dc32

  Copy full SHA for 333dc32

Commits on Dec 30, 2016

• Remove so many more hourglass imports

  Show description for e121667
  

  jart

  authored and

  tensorflower-gardener

  committed
  e121667

  Copy full SHA for e121667

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 1, 2016

• Move the util classes out of sdca_ops.py.

  Show description for d5c9d24
  tensorflower-gardener

  committed
  d5c9d24

  Copy full SHA for d5c9d24

Commits on Nov 22, 2016

• No longer expose the as_ref argument from convert_to_tensor and friends.

  Show description for 0672d41
  tensorflower-gardener

  committed
  0672d41

  Copy full SHA for 0672d41

Commits on Oct 31, 2016

• Migrate tf.learn to use the new summary ops (tf.summary.X).

  Show description for 1962804
  

  Dan Mané

  authored and

  tensorflower-gardener

  committed
  1962804

  Copy full SHA for 1962804

Commits on Oct 28, 2016

• Change SDCA to use DenseMutableHashtable.

  Show description for 87a61b1
  tensorflower-gardener

  committed
  87a61b1

  Copy full SHA for 87a61b1

Commits on Oct 19, 2016

• In tf.contrib, only load external op libraries on non-Windows platforms.

  Show description for 45c8386
  

  mrry

  authored and

  tensorflower-gardener

  committed
  45c8386

  Copy full SHA for 45c8386

Commits on Oct 12, 2016

• Make updating weights optional in SDCA.

  Show description for fe84821
  tensorflower-gardener

  committed
  fe84821

  Copy full SHA for fe84821

Commits on Sep 29, 2016

• Fix broken lists in documentation, and back-quote more identifiers.

  Show description for dd7d5dc
  

  Patrick Nguyen

  authored and

  tensorflower-gardener

  committed
  dd7d5dc

  Copy full SHA for dd7d5dc

Commits on Sep 27, 2016

• Move SDCA optimizer kernels to core.

  Show description for eb4421e
  tensorflower-gardener

  committed
  eb4421e

  Copy full SHA for eb4421e

Commits on Sep 23, 2016

• Add sparse storage support for sparse features for SDCA to support smaller mini batches without any performance penalties.

  Show description for 1cf9120
  tensorflower-gardener

  committed
  1cf9120

  Copy full SHA for 1cf9120

Commits on Sep 16, 2016

• Expose number of partitions to tf.learn SVM Estimator for multi-machine

  Show description for 1c137bd
  tensorflower-gardener

  committed
  1c137bd

  Copy full SHA for 1c137bd

• Smooth hinge loss for SDCA

  Show description for bc85766
  tensorflower-gardener

  committed
  bc85766

  Copy full SHA for bc85766

Commits on Sep 3, 2016

• Add examples_seen summary for the number of distinct examples we have encountered.

  Show description for 01e28de
  tensorflower-gardener

  committed
  01e28de

  Copy full SHA for 01e28de

Commits on Aug 26, 2016

• Make sure the large sum in the duality gap computation happens on the PS

  Show description for 91dc7c9
  tensorflower-gardener

  committed
  91dc7c9

  Copy full SHA for 91dc7c9

Commits on Aug 24, 2016

• - Change float to double which stabilizes the duality gap

  Show description for 698cf5f
  tensorflower-gardener

  committed
  698cf5f

  Copy full SHA for 698cf5f

Commits on Aug 21, 2016

• Add summary for approximate duality gap.

  Show description for 3cc1ef6
  tensorflower-gardener

  committed
  3cc1ef6

  Copy full SHA for 3cc1ef6

Commits on Aug 16, 2016

• Allow num_table_shards to be specified in SDCAOptimizer.

  Show description for f5abc68
  tensorflower-gardener

  committed
  f5abc68

  Copy full SHA for f5abc68

Commits on Aug 13, 2016

• Add num_partitions arg to the SDCA optimizer for distribution and deprecate `container` argument.

  Show description for cd25808
  tensorflower-gardener

  committed
  cd25808

  Copy full SHA for cd25808

• Colocate unshrinked weights with the regular (shrinked) weights and give them explicit names for easier debugging.

  Show description for ccac1bf
  tensorflower-gardener

  committed
  ccac1bf

  Copy full SHA for ccac1bf

Commits on Aug 11, 2016

• Shrink variables independently so that we can assign variables to different parameter servers

  Show description for c955418
  tensorflower-gardener

  committed
  c955418

  Copy full SHA for c955418

Commits on Aug 10, 2016

• Change Saver to support checkpointing ops other than Variables.

  Show description for 0bfd671
  tensorflower-gardener

  committed
  0bfd671

  Copy full SHA for 0bfd671

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 5, 2016

• Support multi-dimensional RealValuedColumns in SDCAOptimizer. Update LinearClassifier and SVMs test to excercise this change.

  Show description for 27a4509
  tensorflower-gardener

  committed
  27a4509

  Copy full SHA for 27a4509

Commits on Jul 19, 2016

• Rewrite to make SDCA optimizer distributed.

  Show description for 6e6d0dc
  tensorflower-gardener

  committed
  6e6d0dc

  Copy full SHA for 6e6d0dc

Commits on Jul 9, 2016

• Add ShardedHashTable private to SDCA in preparation for distributed training.

  Show description for 5a8aac0
  tensorflower-gardener

  committed
  5a8aac0

  Copy full SHA for 5a8aac0

Commits on Jun 30, 2016

• Cleanup SDCA op based on additional review comments.

  Show description for 01fb3ef
  tensorflower-gardener

  committed
  01fb3ef

  Copy full SHA for 01fb3ef

Commits on Jun 28, 2016

• Slightly simplifies hinge_loss computation using nn.relu

  Show description for 055fdda
  tensorflower-gardener

  committed
  055fdda

  Copy full SHA for 055fdda

Commits on Jun 27, 2016

• Change SDCA to use MutableHashTable op instead of internal hash table.

  Show description for 370a6d4
  tensorflower-gardener

  committed
  370a6d4

  Copy full SHA for 370a6d4

Commits on Jun 2, 2016

• Update copyright for 3p/tf.

  Show description for 122cdce
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  122cdce

  Copy full SHA for 122cdce

Commits on Apr 28, 2016

• TensorFlow: colocate_with is unnecessary for inputs that take refs

  Show description for 8cf1568
  

  Vijay Vasudevan

  authored and

  tensorflower-gardener

  committed
  8cf1568

  Copy full SHA for 8cf1568

Commits on Apr 14, 2016

• Removing lazy loading of sdca_ops.so since it is no longer necessary.

  Show description for c25a3fa
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  c25a3fa

  Copy full SHA for c25a3fa

Commits on Apr 12, 2016

• Removing lazy loading of sdca_ops.so since it is no longer necessary.

  Show description for f5258ed
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  f5258ed

  Copy full SHA for f5258ed

• Removing lazy loading of sdca_ops.so since it is no longer necessary.

  Show description for ca9cfc9
  

  A. Unique TensorFlower

  authored and

  tensorflower-gardener

  committed
  ca9cfc9

  Copy full SHA for ca9cfc9