Commits

Branch selector

 master

User selector

fchollet

Datepicker

All time

Commit History

Commits on Jun 12, 2018

• Minor fixes in tf.keras codebase in preparation for Keras 2.2.0 API support.

  Show description for abfdf45
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  abfdf45

  Copy full SHA for abfdf45

Commits on Jun 8, 2018

• Preserve input shape information when serializing deferred-build Sequential models.

  Show description for 255a1c4
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  255a1c4

  Copy full SHA for 255a1c4

Commits on Jun 7, 2018

• Download tf.keras datasets from GCS and add license information.

  Show description for f6ead21
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  f6ead21

  Copy full SHA for f6ead21

Commits on Jun 4, 2018

• Add various missing aliases for symbols in tf.keras submodules.

  Show description for 6b2a088
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  6b2a088

  Copy full SHA for 6b2a088

Commits on May 29, 2018

• Add tf.keras programmer's guide.

  Show description for 5f9f3c7
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  5f9f3c7

  Copy full SHA for 5f9f3c7

Commits on May 24, 2018

• Raise ValueError when calling model.summary() before it is built

  Show description for b42bc6e
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  b42bc6e

  Copy full SHA for b42bc6e

• Add shape validation for symbolic tensors passed to fit (only graph mode).

  Show description for 677b4cf
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  677b4cf

  Copy full SHA for 677b4cf

Commits on May 11, 2018

• Partial update of tf.keras to the Keras 2.1.6 API.

  Show description for cf4cc85
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  cf4cc85

  Copy full SHA for cf4cc85

Commits on May 10, 2018

• Enable Model training/eval from generator in eager execution. Fixes #18287

  Show description for 1b67ccb
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  1b67ccb

  Copy full SHA for 1b67ccb

• Fix inaccurate docstring of Orthogonal initializer.

  Show description for af4cd0e
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  af4cd0e

  Copy full SHA for af4cd0e

Commits on May 9, 2018

• Fix bug in which the ConvLSTM2D layer could not be cloned.

  Show description for cf04e06
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  cf04e06

  Copy full SHA for cf04e06

Commits on May 3, 2018

• Allow `Layer.add_loss` to receive non-tensor; fixes error triggered when using a weight regularizer of factor 0.

  Show description for 8a022d3
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  8a022d3

  Copy full SHA for 8a022d3

Commits on Apr 25, 2018

• Fix critical bug in metrics computation in Eager with Model.

  Show description for 44c9c3a
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  44c9c3a

  Copy full SHA for 44c9c3a

Commits on Apr 24, 2018

• Small refactor of tf.keras aiming at centralizing reusable utilities in `utils`.

  Show description for de3e983
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  de3e983

  Copy full SHA for de3e983

Commits on Apr 23, 2018

• Add `static_state_saving_rnn` back to the `nn` module.

  Show description for 5809ad4
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  5809ad4

  Copy full SHA for 5809ad4

Commits on Apr 20, 2018

• Fix loss computation bug in Model training/eval methods with eager execution enabled.

  Show description for a722cdf
  fchollet

  committed
  a722cdf

  Copy full SHA for a722cdf

Commits on Apr 19, 2018

• Add support for Dataset Iterators in Model training/eval methods in graph mode.

  Show description for 16d25e8
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  16d25e8

  Copy full SHA for 16d25e8

Commits on Apr 18, 2018

• Fix loss computation bug in Model training/eval methods with eager execution enabled.

  Show description for 5c1e253
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  5c1e253

  Copy full SHA for 5c1e253

Commits on Apr 13, 2018

• Avoid mixing `Dimension` type and `int` when defining kernel shapes in conv layers.

  Show description for 026f052
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  026f052

  Copy full SHA for 026f052

• Extend Keras symbol-feeding to dynamic-length tensors and tensors of different dtypes from the target placeholders.

  Show description for 92f870d
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  92f870d

  Copy full SHA for 92f870d

• Enable efficient feeding of symbolic tensors to placeholders in the Keras backend.

  Show description for 5a6d5a1
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  5a6d5a1

  Copy full SHA for 5a6d5a1

• Avoid calling K.learning_phase() when not necessary in Dropout layer since it instantiates a placeholder_with_default, which is not supported by TPU compilation.

  Show description for c4526e5
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  c4526e5

  Copy full SHA for c4526e5

Commits on Apr 11, 2018

• When not necessary, avoid the creation of a `placeholder_with_default` in BN (not yet supported by TPU compilation).

  Show description for 744a5cc
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  744a5cc

  Copy full SHA for 744a5cc

• Remove BN workaround for resource variable gradients bug that was recently fixed.

  Show description for 963ad0f
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  963ad0f

  Copy full SHA for 963ad0f

• Fix for users who were passing `Dimension` type as `units` arg in `Dense`.

  Show description for fad7478
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  fad7478

  Copy full SHA for fad7478

Commits on Apr 10, 2018

• Fix `nn` module RNN namespace issues.

  Show description for dc8aa01
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  dc8aa01

  Copy full SHA for dc8aa01

• Refactor layers:

  Show description for 693b339
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  693b339

  Copy full SHA for 693b339

Commits on Apr 3, 2018

• Add ability to pass symbolic tensors as inputs and targets in calls to Model training and evaluation methods.

  Show description for 1948b74
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  1948b74

  Copy full SHA for 1948b74

Commits on Mar 25, 2018

• Trivial update of layer imports in eager execution examples, to reflect recommended practices.

  Show description for f95347a
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  f95347a

  Copy full SHA for f95347a

Commits on Mar 23, 2018

• Refactor keras.Sequential and enable building Sequential models without specifying an input shape in advance (the model gets built on fit/etc, like subclassed models).

  Show description for 8038376
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  8038376

  Copy full SHA for 8038376

Commits on Mar 17, 2018

• Merge pull request #17668 from sibyjackgrove/patch-2

  Show description for 48178e0
  fchollet

  authored
  48178e0

  Copy full SHA for 48178e0

Commits on Mar 9, 2018

• Part of the update of tf.keras to the Keras 2.1.5 API.

  Show description for 8044288
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  8044288

  Copy full SHA for 8044288

Commits on Mar 6, 2018

• Introduce API `keras.backend.learning_phase_scope(value)` (internal for now).

  Show description for cebb7fc
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  cebb7fc

  Copy full SHA for cebb7fc

• Fixes a number of usability issues with model_to_estimator, in particular:

  Show description for 73999dc
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  73999dc

  Copy full SHA for 73999dc

Commits on Mar 5, 2018

• Fix bug with multi_gpu_model / model.inputs.

  Show description for 1e39064
  

  fchollet

  authored and

  tensorflower-gardener

  committed
  1e39064

  Copy full SHA for 1e39064

Pagination

Previous

Next