Commits · d51cc280f68a8bfb4b1b360248ea592aa71f845a · ModelZoo / ResNet50_tensorflow

13 Apr, 2022 1 commit
- Internal change · d51cc280
  Jialu Liu authored Apr 12, 2022
```
PiperOrigin-RevId: 441337748
```
  d51cc280
30 Mar, 2022 1 commit
- Updated the text and added some formatting to render better. · 82949522
  A. Unique TensorFlower authored Mar 29, 2022
```
PiperOrigin-RevId: 438209905
```
  82949522
28 Mar, 2022 1 commit
- Internal change · ec55d7da
  Luke Wood authored Mar 28, 2022
```
PiperOrigin-RevId: 437812893
```
  ec55d7da
23 Mar, 2022 2 commits
- Internal change · c166ae02
  Yilei Yang authored Mar 23, 2022
```
PiperOrigin-RevId: 436824079
```
  c166ae02
- Internal change · 244b66bb
  Jiayu Ye authored Mar 22, 2022
```
PiperOrigin-RevId: 436647758
```
  244b66bb
11 Mar, 2022 5 commits
- Fix typo. · ca17d835
  Hongkun Yu authored Mar 11, 2022
```
PiperOrigin-RevId: 434053895
```
  ca17d835
- Internal change · 3ab3dd6a
  A. Unique TensorFlower authored Mar 11, 2022
```
PiperOrigin-RevId: 434027325
```
  3ab3dd6a
- Change BertPretrainer to keep `mlm_activation` in config. · 68039be4
  Jinoo Baek authored Mar 11, 2022
```
PiperOrigin-RevId: 434014162
```
  68039be4
- Internal change · 8f815365
  Le Hou authored Mar 10, 2022
```
PiperOrigin-RevId: 433920851
```
  8f815365
- Internal change · 7b2df02f
  Jinoo Baek authored Mar 10, 2022
```
PiperOrigin-RevId: 433892706
```
  7b2df02f
07 Mar, 2022 1 commit
- Internal change · fd424b07
  Jiayu Ye authored Mar 07, 2022
```
PiperOrigin-RevId: 432997436
```
  fd424b07
04 Mar, 2022 1 commit
- Internal change · 71a6b899
  A. Unique TensorFlower authored Mar 03, 2022
```
PiperOrigin-RevId: 432311587
```
  71a6b899
03 Mar, 2022 1 commit
- Internal change · b136d845
  A. Unique TensorFlower authored Mar 03, 2022
```
PiperOrigin-RevId: 432292375
```
  b136d845
01 Mar, 2022 1 commit
- Internal change · 10ee28dd
  Jiayu Ye authored Mar 01, 2022
```
PiperOrigin-RevId: 431722553
```
  10ee28dd
24 Feb, 2022 1 commit
- Internal change · a0b548e2
  Jiayu Ye authored Feb 24, 2022
```
PiperOrigin-RevId: 430740048
```
  a0b548e2
09 Feb, 2022 1 commit
- Internal change · d99b9407
  Jiayu Ye authored Feb 09, 2022
```
PiperOrigin-RevId: 427509556
```
  d99b9407
25 Jan, 2022 1 commit
- Internal change · 885fda09
  A. Unique TensorFlower authored Jan 25, 2022
```
PiperOrigin-RevId: 424187808
```
  885fda09
19 Jan, 2022 1 commit
- Internal change · a8bccfa4
  A. Unique TensorFlower authored Jan 19, 2022
```
PiperOrigin-RevId: 422917125
```
  a8bccfa4
18 Jan, 2022 1 commit
- Internal change · 871c4e0a
  A. Unique TensorFlower authored Jan 18, 2022
```
PiperOrigin-RevId: 422612138
```
  871c4e0a
07 Jan, 2022 1 commit
- Internal change · 993dbf54
  Jiayu Ye authored Jan 07, 2022
```
PiperOrigin-RevId: 420368411
```
  993dbf54
04 Jan, 2022 1 commit
- [reuse] Fix Order-dependent test. The root case is that large input data also increases variance. · 9f3443f9
  Frederick Liu authored Jan 04, 2022
```
PiperOrigin-RevId: 419617435
```
  9f3443f9
29 Dec, 2021 1 commit

Added block diagonal feedforward layer. · a7e60974

Xin Wang authored Dec 29, 2021

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

a7e60974

21 Dec, 2021 1 commit
- Internal change · 446da419
  Frederick Liu authored Dec 21, 2021
```
PiperOrigin-RevId: 417646384
```
  446da419
16 Dec, 2021 1 commit
- Release the fast tokenizer bert wrapper on github: https://arxiv.org/abs/2012.15524 · 782e39e8
  Hongkun Yu authored Dec 15, 2021
```
PiperOrigin-RevId: 416671104
```
  782e39e8
03 Dec, 2021 2 commits
- Fix activation hinting. · 46d6af72
  Yuexin Wu authored Dec 03, 2021
```
PiperOrigin-RevId: 414032298
```
  46d6af72
- Allow Funnel Transformer to switch between basic TransformerBlocks (added ReZero). · b493efb2
  Yuexin Wu authored Dec 03, 2021
```
PiperOrigin-RevId: 414027254
```
  b493efb2
30 Nov, 2021 1 commit
- Allow ReZero to take 3 inputs that can be used for compressing sequence length. · 6af14369
  Yuexin Wu authored Nov 30, 2021
```
PiperOrigin-RevId: 413261350
```
  6af14369
25 Nov, 2021 1 commit
- Internal change · 68e59e15
  Hongkun Yu authored Nov 24, 2021
```
PiperOrigin-RevId: 412185559
```
  68e59e15
23 Nov, 2021 1 commit
- Internal change · f6557386
  Frederick Liu authored Nov 22, 2021
```
PiperOrigin-RevId: 411729044
```
  f6557386
19 Nov, 2021 1 commit
- Internal change · 19e3d234
  Frederick Liu authored Nov 19, 2021
```
PiperOrigin-RevId: 411088753
```
  19e3d234
10 Nov, 2021 4 commits
- Internal change · 79b6de8e
  Frederick Liu authored Nov 10, 2021
```
PiperOrigin-RevId: 408977858
```
  79b6de8e
- Internal change · d872cee2
  Frederick Liu authored Nov 10, 2021
```
PiperOrigin-RevId: 408970557
```
  d872cee2
- [reuse] Add layers used in [Leveraging redundancy in attention with Reuse... · 892dac23
  Frederick Liu authored Nov 10, 2021
```
[reuse] Add layers used in [Leveraging redundancy in attention with Reuse Transformers](https://arxiv.org/abs/2110.06821).

PiperOrigin-RevId: 408969659
```
  892dac23
- Remove @tf.keras.utils.register_keras_serializable(package="Text") as the... · a241b9ae
  Hongkun Yu authored Nov 10, 2021
```
Remove @tf.keras.utils.register_keras_serializable(package="Text") as the model cannot be serialized anyway.

PiperOrigin-RevId: 408904971
```
  a241b9ae
09 Nov, 2021 2 commits
- Internal change · 7f2c1d5a
  A. Unique TensorFlower authored Nov 09, 2021
```
PiperOrigin-RevId: 408650343
```
  7f2c1d5a
- Internal change · 46c9fd72
  A. Unique TensorFlower authored Nov 08, 2021
```
PiperOrigin-RevId: 408473818
```
  46c9fd72
04 Nov, 2021 1 commit
- Internal change · a80c26b1
  A. Unique TensorFlower authored Nov 04, 2021
```
PiperOrigin-RevId: 407664875
```
  a80c26b1
03 Nov, 2021 1 commit
- #Funnel fix `mixed_precision=None` error when mixed with TF1 code. · fdd0a8a3
  Yuexin Wu authored Nov 02, 2021
```
PiperOrigin-RevId: 407253800
```
  fdd0a8a3
02 Nov, 2021 1 commit

[bert_encoder] Found a typo in previous experiment and turns out that the... · f32dea32

Frederick Liu authored Nov 01, 2021

[bert_encoder] Found a typo in previous experiment and turns out that the checkpoints are compatible even when wrapped in models.

PiperOrigin-RevId: 406987505

f32dea32

01 Nov, 2021 1 commit
- Internal change · ab70b0f2
  A. Unique TensorFlower authored Nov 01, 2021
```
PiperOrigin-RevId: 406906304
```
  ab70b0f2