提交 · 4e9f9514cbbd623f37ec521afdafc8e10496f0db · 曾经的那一瞬间 / Models

03 8月, 2022 1 次提交
- H
  Internal change · 4e9f9514
  由 Hongkun Yu 提交于 8月 02, 2022
```
PiperOrigin-RevId: 464934071
```
  4e9f9514
04 5月, 2022 1 次提交
- S
  Internal change · 6d458bcc
  由 Scott Zhu 提交于 5月 03, 2022
```
PiperOrigin-RevId: 446219773
```
  6d458bcc
19 1月, 2022 1 次提交
- A
  Internal change · 871c4e0a
  由 A. Unique TensorFlower 提交于 1月 18, 2022
```
PiperOrigin-RevId: 422612138
```
  871c4e0a
11 11月, 2021 1 次提交
- F
  Internal change · d872cee2
  由 Frederick Liu 提交于 11月 10, 2021
```
PiperOrigin-RevId: 408970557
```
  d872cee2
08 10月, 2021 1 次提交
- F
  [keras_nlp] Merge keras_nlp into tf_nlp. · 9b23daf9
  由 Frederick Liu 提交于 10月 07, 2021
```
PiperOrigin-RevId: 401593694
```
  9b23daf9
25 8月, 2021 1 次提交
- A
  Internal change · 0b098c60
  由 A. Unique TensorFlower 提交于 8月 25, 2021
```
PiperOrigin-RevId: 392901555
```
  0b098c60
18 5月, 2021 1 次提交
- F
  Internal change · 4f5ad013
  由 Frederick Liu 提交于 5月 17, 2021
```
PiperOrigin-RevId: 374236491
```
  4f5ad013
24 4月, 2021 1 次提交
- C
  Internal change · 68ee72ae
  由 Chen Chen 提交于 4月 23, 2021
```
PiperOrigin-RevId: 370159521
```
  68ee72ae
13 3月, 2021 1 次提交
- C
  Use backticks to denote code spans in nlp modeling docstrings · 94dd4911
  由 Chen Chen 提交于 3月 12, 2021
```
PiperOrigin-RevId: 362610475
```
  94dd4911
10 3月, 2021 1 次提交
- F
  Internal change · 5a898973
  由 Frederick Liu 提交于 3月 09, 2021
```
PiperOrigin-RevId: 361957289
```
  5a898973
19 2月, 2021 1 次提交
- A
  Remove a redundant line because it is the same as line 242. · aeab8791
  由 A. Unique TensorFlower 提交于 2月 18, 2021
```
PiperOrigin-RevId: 358301569
```
  aeab8791
21 12月, 2020 1 次提交
- S
  
  [official/**.py] Rename "Arguments:" to "Args:" · 90979a21
  由 Samuel Marks 提交于 12月 21, 2020
  
  90979a21
10 12月, 2020 1 次提交
- H
  Consolidate the self attention mask layer. · 5fc4c351
  由 Hongkun Yu 提交于 12月 09, 2020
```
Ask users to use keras_nlp.

PiperOrigin-RevId: 346621226
```
  5fc4c351
19 11月, 2020 2 次提交
- L
  minor bug fix · 16ca3721
  由 Le Hou 提交于 11月 18, 2020
```
PiperOrigin-RevId: 343154699
```
  16ca3721
- L
  Support adding a layer normalization layer before the final pooling layer. · 58edfb5c
  由 Le Hou 提交于 11月 18, 2020
```
PiperOrigin-RevId: 343092077
```
  58edfb5c
28 10月, 2020 1 次提交
- L
  Save layers for access in EncoderScaffold. · 9f9b4307
  由 Le Hou 提交于 10月 27, 2020
```
PiperOrigin-RevId: 339345258
```
  9f9b4307
27 10月, 2020 1 次提交
- H
  Py3 cleanup for nlp/ · cd5e1204
  由 Hongkun Yu 提交于 10月 26, 2020
```
PiperOrigin-RevId: 339071563
```
  cd5e1204
24 10月, 2020 1 次提交

Shift all attribute assignments in functional subclass models below the... · 589f0269

由 A. Unique TensorFlower 提交于 10月 23, 2020

Shift all attribute assignments in functional subclass models below the super().__init__ call, and enable setattr tracking on all functional subclasses.

PiperOrigin-RevId: 338759531

589f0269

14 10月, 2020 1 次提交
- A
  Slice Transformer outputs with a TF Op layer instead of a Lambda layer. · cd39fd58
  由 A. Unique TensorFlower 提交于 10月 13, 2020
```
Lambda layers are fundamentally non-portable since they serialize
Python bytecode.

PiperOrigin-RevId: 336917172
```
  cd39fd58
12 9月, 2020 1 次提交
- Z
  aliasing OnDeviceEmbedding inside tensorflow_models. · 1a273519
  由 Zhenyu Tan 提交于 9月 11, 2020
```
PiperOrigin-RevId: 331173006
```
  1a273519
10 9月, 2020 1 次提交
- H
  Add dict outputs to BertEncoder and EncoderScaffold · f1b9d43f
  由 Hongkun Yu 提交于 9月 09, 2020
```
PiperOrigin-RevId: 330817736
```
  f1b9d43f
02 9月, 2020 1 次提交
- Z
  Add PositionEmbedding for keras_nlp. · 1e22d1fe
  由 Zhenyu Tan 提交于 9月 01, 2020
```
PiperOrigin-RevId: 329568227
```
  1e22d1fe
26 8月, 2020 2 次提交
- A
  Get the handle of self._embedding_norm_layer for future weight copying. · f42b8392
  由 A. Unique TensorFlower 提交于 8月 25, 2020
```
PiperOrigin-RevId: 328389991
```
  f42b8392
- A
  Get the handle of self._embedding_norm_layer for future weight copying. · a1a7aaca
  由 A. Unique TensorFlower 提交于 8月 25, 2020
```
PiperOrigin-RevId: 328389991
```
  a1a7aaca
20 8月, 2020 2 次提交
- A
  Log encoder scaffold arguments. · 1704f984
  由 A. Unique TensorFlower 提交于 8月 19, 2020
```
PiperOrigin-RevId: 327518640
```
  1704f984
- A
  Log encoder scaffold arguments. · 6b2bc083
  由 A. Unique TensorFlower 提交于 8月 19, 2020
```
PiperOrigin-RevId: 327518640
```
  6b2bc083
18 8月, 2020 2 次提交
- A
  Minor change in EncoderScaffold · a0342123
  由 A. Unique TensorFlower 提交于 8月 17, 2020
```
PiperOrigin-RevId: 327160430
```
  a0342123
- A
  Minor change in EncoderScaffold · 1b5b2fe4
  由 A. Unique TensorFlower 提交于 8月 17, 2020
```
PiperOrigin-RevId: 327160430
```
  1b5b2fe4
13 8月, 2020 2 次提交
- H
  Internal change · 999fae62
  由 Hongkun Yu 提交于 8月 12, 2020
```
PiperOrigin-RevId: 326286926
```
  999fae62
- H
  Internal change · 88253ce5
  由 Hongkun Yu 提交于 8月 12, 2020
```
PiperOrigin-RevId: 326286926
```
  88253ce5
31 7月, 2020 2 次提交
- H
  Internal change · afee40a0
  由 Hongkun Yu 提交于 7月 30, 2020
```
PiperOrigin-RevId: 324137297
```
  afee40a0
- H
  Internal change · a62c2bfc
  由 Hongkun Yu 提交于 7月 30, 2020
```
PiperOrigin-RevId: 324137297
```
  a62c2bfc
17 7月, 2020 2 次提交
- H
  Adds clear documentation: Functional/Subclass API used for each network/model. · 597e59d2
  由 Hongkun Yu 提交于 7月 16, 2020
```
PiperOrigin-RevId: 321591514
```
  597e59d2
- H
  Adds clear documentation: Functional/Subclass API used for each network/model. · 7e47cd7b
  由 Hongkun Yu 提交于 7月 16, 2020
```
PiperOrigin-RevId: 321591514
```
  7e47cd7b
15 5月, 2020 1 次提交
- C
  Internal change · 8c408bbe
  由 Chen Chen 提交于 5月 14, 2020
```
PiperOrigin-RevId: 311597242
```
  8c408bbe
14 5月, 2020 1 次提交
- C
  Internal change · 3c5d118c
  由 Chen Chen 提交于 5月 13, 2020
```
PiperOrigin-RevId: 311428193
```
  3c5d118c
13 5月, 2020 1 次提交

Restructure the Keras class hierarchy for Network, Model and Sequential. · d08af90d

由 Scott Zhu 提交于 5月 12, 2020

The intention of this change is to reduce the code complexity within Keras class, especially for Network, which currently contains logic for both subclass Model and functional Model.

After this change, the subclass model and functional model become individual class and become self contained.

1. Model is now the base class for subclass model. It doesn't contains network structure management, and the topology will be created within __init__ and __call__, which is for user to implement. It also contains compile/fit/eval/predict, which is the basic functionality for model training.

2. Functional is created based on existing Network class. It extends the Model, which allows it leverage compile/fit/eval/predict. In addition, it also take input/output as init parameter and manage the network topology.

3. Sequential model is now a subclass of Functional, since it will use Functional's method to manage it topology (layer stacking).

Model(input, output) will create a Functional under the hood, and behave the same way as before.

PiperOrigin-RevId: 311232972

d08af90d

09 5月, 2020 1 次提交

Add an interface in nlp.modeling.networks.encoder_scaffold.EncoderScaffold and... · 5e641c43

由 A. Unique TensorFlower 提交于 5月 08, 2020

Add an interface in nlp.modeling.networks.encoder_scaffold.EncoderScaffold and nlp.modeling.networks.transformer_encoder.TransformerEncoder, to get the pooler Dense layer reference.

PiperOrigin-RevId: 310675150

5e641c43

08 5月, 2020 1 次提交

Add interfaces in nlp.modeling.networks.encoder_scaffold.EncoderScaffold,... · 56a2f704

由 A. Unique TensorFlower 提交于 5月 07, 2020

Add interfaces in nlp.modeling.networks.encoder_scaffold.EncoderScaffold, allowing it to output all hidden layer references, and all intermediate output data references.

PiperOrigin-RevId: 310509202

56a2f704

21 4月, 2020 1 次提交
- C
  Set the name when creating layers.PositionEmbedding object. · 1f685c54
  由 Chen Chen 提交于 4月 20, 2020
```
PiperOrigin-RevId: 307500045
```
  1f685c54

曾经的那一瞬间 / Models 大约 1 年 前同步成功

曾经的那一瞬间 / Models
大约 1 年前同步成功