Commits

Branch selector

 master

User selector

haozha111

Datepicker

All time

Commit History

Commits on Nov 10, 2021

• Remove a TODO

  Show description for 5bd6a7e
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  5bd6a7e

  Copy full SHA for 5bd6a7e

• In analyze variables pass, make the following changes:

  Show description for 45a8dad
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  45a8dad

  Copy full SHA for 45a8dad

Commits on Nov 9, 2021

• Improve on-device training colab.

  Show description for 222970e
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  222970e

  Copy full SHA for 222970e

• Improve on-device training colab.

  Show description for 2706087
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  2706087

  Copy full SHA for 2706087

Commits on Nov 8, 2021

• Add TODO about resource ID generation.

  Show description for 4d67711
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  4d67711

  Copy full SHA for 4d67711

• Fix build dependency for training delegate.

  Show description for 41424fd
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  41424fd

  Copy full SHA for 41424fd

Commits on Nov 5, 2021

• Remove selective registration on tflite training delegate.

  Show description for 607480d
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  607480d

  Copy full SHA for 607480d

Commits on Nov 4, 2021

• Add a helper function IsResourceOrVariant.

  Show description for a705d76
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  a705d76

  Copy full SHA for a705d76

• Remove false assumption in delegate test code.

  Show description for 9c32976
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  9c32976

  Copy full SHA for 9c32976

Commits on Nov 2, 2021

• Check status of AllocateTensors in TfLiteSubgraphExecute op.

  Show description for 026a692
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  026a692

  Copy full SHA for 026a692

Commits on Nov 1, 2021

• Separate out common code to share with micro benchmark.

  Show description for c8702ad
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  c8702ad

  Copy full SHA for c8702ad

• Check if tensor's ranked before getting its shape.

  Show description for 3eba5c1
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  3eba5c1

  Copy full SHA for 3eba5c1

• Add a new benchmark flag release_dynamic_tensor to control whether dynamic tensor should be released during benchmark runs.

  Show description for c966787
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  c966787

  Copy full SHA for c966787

Commits on Oct 29, 2021

• Add TakeWhileDataset into flex op list.

  Show description for bb1f2ca
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  bb1f2ca

  Copy full SHA for bb1f2ca

• Clean up and unify mobile build rules for TF data.

  Show description for 1186a60
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  1186a60

  Copy full SHA for 1186a60

Commits on Oct 28, 2021

• Optimize AddN kernel using multi-threading.

  Show description for df5dda8
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  df5dda8

  Copy full SHA for df5dda8

Commits on Oct 27, 2021

• Add more checking in visualize.py to avoid enumerating over NoneTypes.

  Show description for 4731fd6
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  4731fd6

  Copy full SHA for 4731fd6

Commits on Oct 26, 2021

• Allow users to specify dynamic tensors to be released via Interpreter.

  Show description for d24b2e9
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  d24b2e9

  Copy full SHA for d24b2e9

Commits on Oct 21, 2021

• Refactor out common code into add_n_test_common.h for easy sharing with op micro-benchmarks.

  Show description for 0749201
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  0749201

  Copy full SHA for 0749201

Commits on Oct 20, 2021

• Change batch size to 100 in on-device training colab.

  Show description for 4b593da
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  4b593da

  Copy full SHA for 4b593da

Commits on Oct 19, 2021

• For train function, use a batched input (default size: 1000) instead of feeding the entire training data. Update documentation as well.

  Show description for 8b1534b
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  8b1534b

  Copy full SHA for 8b1534b

• Add plot graph for training loss in colab.

  Show description for 2fd7398
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  2fd7398

  Copy full SHA for 2fd7398

• Update the on-device training colab with more user-friendly wording and explanation.

  Show description for 36ce694
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  36ce694

  Copy full SHA for 36ce694

Commits on Oct 7, 2021

• Update on-device training colab with the following improvements:

  Show description for 9cfb4b1
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  9cfb4b1

  Copy full SHA for 9cfb4b1

Commits on Sep 17, 2021

• Add more argument type checking for tfl.while.

  Show description for 28d15bc
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  28d15bc

  Copy full SHA for 28d15bc

Commits on Sep 13, 2021

• Add shape inference for Restore and RestoreV2 op.

  Show description for 1946bbc
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  1946bbc

  Copy full SHA for 1946bbc

• Add shape inference for Restore and RestoreV2 op.

  Show description for 8ac4996
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  8ac4996

  Copy full SHA for 8ac4996

Commits on Sep 11, 2021

• Only set the output tenosr of ReadVariable to dynamic if the tensor's dimension size is 0.

  Show description for 0e66a5b
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  0e66a5b

  Copy full SHA for 0e66a5b

Commits on Sep 10, 2021

• Add shape inference for Restore and RestoreV2 op.

  Show description for f62f5ad
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  f62f5ad

  Copy full SHA for f62f5ad

• Add a folding pattern for AddN op, when there is one single non-zero input.

  Show description for 4971f71
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  4971f71

  Copy full SHA for 4971f71

Commits on Sep 3, 2021

• Add FilterDataset into the flex op allowlist and TF portable extended ops.

  Show description for db4931d
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  db4931d

  Copy full SHA for db4931d

Commits on Aug 12, 2021

• Remove dead TODO. When multi-signature API support is added, when calling the GetSubgraphs api, it will still return all the subgraphs contained in the TF Lite model.

  Show description for c6175ea
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  c6175ea

  Copy full SHA for c6175ea

• Remove dead TODO.

  Show description for fc6619c
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  fc6619c

  Copy full SHA for fc6619c

Commits on Aug 11, 2021

• Remove TODO since multiple entry points is supported.

  Show description for 73b2a99
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  73b2a99

  Copy full SHA for 73b2a99

• Remove an old TODO associated with TOCO (since TOCO has already been deprecated).

  Show description for 9d97737
  

  haozha111

  authored and

  tensorflower-gardener

  committed
  9d97737

  Copy full SHA for 9d97737

Pagination

Previous

Next