Commits

Branch selector

 master

User selector

pavithrasv

Datepicker

All time

Commit History

Commits on Sep 13, 2019

• Fix performance regression issue by reusing metrics property.

  Show description for ebc4653
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  ebc4653

  Copy full SHA for ebc4653

Commits on Sep 4, 2019

• Fix add_metric name ordering issue.

  Show description for 92ac5c7
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  92ac5c7

  Copy full SHA for 92ac5c7

Commits on Sep 3, 2019

• Automated rollback of commit c3fb862245d7d3daa318f3e04de25fed83d29993

  Show description for e3a3d33
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  e3a3d33

  Copy full SHA for e3a3d33

Commits on Aug 31, 2019

• Automated rollback of commit e75cf2b3c359332d8ec200358eb2a86fddadf2ed

  Show description for c3fb862
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  c3fb862

  Copy full SHA for c3fb862

• Negating cosine similarity loss so that value gets minimized while training without needed a wrapper function.

  Show description for e75cf2b
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  e75cf2b

  Copy full SHA for e75cf2b

Commits on Aug 30, 2019

• Raise error when sparse ops are used with TensorFlowOpLayer.

  Show description for e3805e7
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  e3805e7

  Copy full SHA for e3805e7

Commits on Aug 22, 2019

• Add support for sample weight mode in v2 single execution path.

  Show description for 9f30dc6
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  9f30dc6

  Copy full SHA for 9f30dc6

• Updating regex in test case to a more stable one with inline dotall flag.

  Show description for 6fd8280
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  6fd8280

  Copy full SHA for 6fd8280

Commits on Aug 21, 2019

• 1. Fix incorrect steps inference when validation_split is provided in fit in v2 single path execution.

  Show description for 4607de4
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  4607de4

  Copy full SHA for 4607de4

Commits on Aug 20, 2019

• Enable single execution path for `weighted_metrics` param in compile.

  Show description for 130a306
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  130a306

  Copy full SHA for 130a306

Commits on Aug 15, 2019

• Fixing validation callback configuration in fit in single execution path.

  Show description for 2009d21
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  2009d21

  Copy full SHA for 2009d21

Commits on Aug 14, 2019

• 1. Do not raise `steps unsupported with numpy arrays` warning message in single execution path.

  Show description for a59ad83
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  a59ad83

  Copy full SHA for a59ad83

Commits on Aug 3, 2019

• Enable the Keras compile API `experimental_run_tf_function` flag by default.

  Show description for 54569bd
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  54569bd

  Copy full SHA for 54569bd

Commits on Aug 2, 2019

• Remove Sequence condition check that is around input casting.

  Show description for dad0d6a
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  dad0d6a

  Copy full SHA for dad0d6a

Commits on Aug 1, 2019

• Fix count params function and add unit test.

  Show description for 40e694d
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  40e694d

  Copy full SHA for 40e694d

• Increasing shard count of temporal sample weights test to fix timeouts.

  Show description for 4157717
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  4157717

  Copy full SHA for 4157717

• Fallback and do not use single execution path if v1 optimizers are used.

  Show description for 469d45c
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  469d45c

  Copy full SHA for 469d45c

• Remove check for partial batch in stateful models if input is a dataset as we do not know the number of samples.

  Show description for beee660
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  beee660

  Copy full SHA for beee660

Commits on Jul 30, 2019

• Renames the `run_distributed` flag to `experimental_run_tf_function` and modifies the tests generated by the keras_all_modes decorator to better reflect that.

  Show description for c7933ce
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  c7933ce

  Copy full SHA for c7933ce

• Add support for partial targets use case in eager mode.

  Show description for ed19de6
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  ed19de6

  Copy full SHA for ed19de6

Commits on Jul 29, 2019

• Clear losses in compile to fix error when interleaving v1 and v2 execution paths.

  Show description for 96c908f
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  96c908f

  Copy full SHA for 96c908f

• Automated rollback of commit a93ef60cdd200449fdcb67a592bbdfb8eb5ff2f0

  Show description for 98ad9da
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  98ad9da

  Copy full SHA for 98ad9da

Commits on Jul 26, 2019

• Fix positional arg in call test failure in single execution path.

  Show description for 7e59af6
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  7e59af6

  Copy full SHA for 7e59af6

Commits on Jul 25, 2019

• Simplify compile API for single execution path and re-compile when required in case we are falling back to v1 loops.

  Show description for a93ef60
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  a93ef60

  Copy full SHA for a93ef60

• Fix autograph comment in base layer.

  Show description for 12a9859
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  12a9859

  Copy full SHA for 12a9859

Commits on Jul 24, 2019

• Handle partial sample weight use case in the single execution path data adapter.

  Show description for c5f031c
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  c5f031c

  Copy full SHA for c5f031c

Commits on Jul 23, 2019

• Raise ValueError if an integer is passed to the training APIs.

  Show description for 63fec20
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  63fec20

  Copy full SHA for 63fec20

Commits on Jul 22, 2019

• Fix invalid `steps` argument usage test for single execution path.

  Show description for e5b12c6
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  e5b12c6

  Copy full SHA for e5b12c6

• 1. Add support for temporal sample weight mode in non-graph networks.

  Show description for 3782019
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  3782019

  Copy full SHA for 3782019

• Add check to see mismatch between input and output batch size in single execution code path.

  Show description for 7341359
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  7341359

  Copy full SHA for 7341359

Commits on Jul 18, 2019

• Automated rollback of commit 11b33344b3b398db10d370c100e915e7ad023546. Revert #30053.

  Show description for e5cf807
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  e5cf807

  Copy full SHA for e5cf807

Commits on Jul 16, 2019

• Fix for scaling when default loss reduction is not specified.

  Show description for 179b388
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  179b388

  Copy full SHA for 179b388

• Improve cross entropy docs to mention why a user should consider from_logits.

  Show description for 42b5da6
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  42b5da6

  Copy full SHA for 42b5da6

Commits on Jul 13, 2019

• Raise error when the expected output batch size of a model does not match the batch size from model output. Fixes #30474.

  Show description for 37fcf0a
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  37fcf0a

  Copy full SHA for 37fcf0a

Commits on Jul 12, 2019

• tf.keras.layers.Layer.count_params() should use _maybe_build() instead of build()

  Show description for 9628781
  

  pavithrasv

  authored and

  tensorflower-gardener

  committed
  9628781

  Copy full SHA for 9628781

Pagination

Previous

Next