- 18 Jan, 2022 3 commits
-
-
Le Hou authored
PiperOrigin-RevId: 422638136
-
Le Hou authored
PiperOrigin-RevId: 422637653
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 422612138
-
- 12 Jan, 2022 1 commit
-
-
Le Hou authored
PiperOrigin-RevId: 421395106
-
- 07 Jan, 2022 1 commit
-
-
Jiayu Ye authored
PiperOrigin-RevId: 420368411
-
- 04 Jan, 2022 8 commits
-
-
Frederick Liu authored
PiperOrigin-RevId: 419638181
-
Frederick Liu authored
PiperOrigin-RevId: 419638181
-
Frederick Liu authored
PiperOrigin-RevId: 419617646
-
Frederick Liu authored
PiperOrigin-RevId: 419617646
-
Frederick Liu authored
PiperOrigin-RevId: 419617435
-
Frederick Liu authored
PiperOrigin-RevId: 419617435
-
Frederick Liu authored
please use Seq2SeqTransformer in tfnlp instead. PiperOrigin-RevId: 419604366
-
Frederick Liu authored
please use Seq2SeqTransformer in tfnlp instead. PiperOrigin-RevId: 419604366
-
- 29 Dec, 2021 6 commits
-
-
Frederick Liu authored
PiperOrigin-RevId: 418851251
-
Frederick Liu authored
PiperOrigin-RevId: 418851251
-
Xin Wang authored
This layer replaces the weight matrix of the output_dense layer with a block diagonal matrix to save layer parameters and FLOPs. A linear mixing layer can be added optionally to improve layer expressibility. PiperOrigin-RevId: 418828099
-
Xin Wang authored
This layer replaces the weight matrix of the output_dense layer with a block diagonal matrix to save layer parameters and FLOPs. A linear mixing layer can be added optionally to improve layer expressibility. PiperOrigin-RevId: 418828099
-
Frederick Liu authored
PiperOrigin-RevId: 418817477
-
Frederick Liu authored
PiperOrigin-RevId: 418817477
-
- 28 Dec, 2021 4 commits
-
-
Frederick Liu authored
PiperOrigin-RevId: 418685487
-
Frederick Liu authored
PiperOrigin-RevId: 418685487
-
Frederick Liu authored
PiperOrigin-RevId: 418559537
-
Frederick Liu authored
PiperOrigin-RevId: 418559537
-
- 21 Dec, 2021 2 commits
-
-
Frederick Liu authored
PiperOrigin-RevId: 417646384
-
Frederick Liu authored
PiperOrigin-RevId: 417646384
-
- 16 Dec, 2021 2 commits
-
-
- 15 Dec, 2021 2 commits
-
-
Hongkun Yu authored
PiperOrigin-RevId: 416450920
-
Hongkun Yu authored
PiperOrigin-RevId: 416450920
-
- 13 Dec, 2021 4 commits
-
-
Hongkun Yu authored
PiperOrigin-RevId: 416088533
-
Hongkun Yu authored
PiperOrigin-RevId: 416088533
-
Hongkun Yu authored
PiperOrigin-RevId: 416070060
-
Hongkun Yu authored
PiperOrigin-RevId: 416070060
-
- 08 Dec, 2021 2 commits
-
-
Smit Hinsu authored
TPUs don't support dynamic Squeeze as it won't be able to identify the dimensions to remove if the input shape is dynamic. For example, [<=16, 1] could return a vector or scalar result depending on the size of the dynamic dim. PiperOrigin-RevId: 415083667
-
Smit Hinsu authored
TPUs don't support dynamic Squeeze as it won't be able to identify the dimensions to remove if the input shape is dynamic. For example, [<=16, 1] could return a vector or scalar result depending on the size of the dynamic dim. PiperOrigin-RevId: 415083667
-
- 07 Dec, 2021 4 commits
-
-
Hongkun Yu authored
PiperOrigin-RevId: 414599442
-
Hongkun Yu authored
PiperOrigin-RevId: 414599442
-
Le Hou authored
PiperOrigin-RevId: 414577439
-
Le Hou authored
PiperOrigin-RevId: 414577439
-
- 03 Dec, 2021 1 commit
-
-
Yuexin Wu authored
PiperOrigin-RevId: 414032298
-