Skip to content

Commit

Permalink
Release v0.2.3
Browse files Browse the repository at this point in the history
* Update version to v0.2.3
* Update CHANGELOG
* Update `examples/text_style_transfer` README
  • Loading branch information
ZhitingHu authored Sep 25, 2019
2 parents 46e006f + f4c04af commit 8dac88d
Show file tree
Hide file tree
Showing 3 changed files with 9 additions and 2 deletions.
7 changes: 6 additions & 1 deletion CHANGELOG.md
Original file line number Diff line number Diff line change
@@ -1,13 +1,18 @@

## [Unreleased]
## [v0.2.3](https://github.com/asyml/texar/releases/tag/v0.2.3) (2019-09-22)

### New features

* Texar Tensorflow now imported by `import texar.tf as tx`. ([#197](https://github.com/asyml/texar/pull/197))
* Add pretrained modules \[[docs](https://texar.readthedocs.io/en/latest/code/modules.html#pre-trained)\], e.g., BERT, XLNet, etc. ([#206](https://github.com/asyml/texar/pull/206))
* Add new tensor shape related utils, e.g., [reduce_with_weights](https://texar.readthedocs.io/en/latest/code/utils.html#reduce-with-weights), [varlength\_\*](https://texar.readthedocs.io/en/latest/code/utils.html#varlength-concat). ([#201](https://github.com/asyml/texar/pull/201))

### Feature improvements

### Fixes

* Fix docstring of connector `_mlp_transform`. ([#192](https://github.com/asyml/texar/pull/192))
* Fix `dynamic_decode` out-of-range error when max-decoding-length is reached. ([#208](https://github.com/asyml/texar/pull/208))

## [v0.2.2](https://github.com/asyml/texar/releases/tag/v0.2.2) (2019-08-05)

Expand Down
2 changes: 2 additions & 0 deletions examples/text_style_transfer/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,8 @@ python main.py --config config
* The model will first be pre-trained for a few epochs (specified in `config.py`). During pre-training, the `Encoder-Decoder` part is trained as an autoencoder, while the `Classifier` part is trained with the classification labels.
* Full-training is then performed for another few epochs. During full-training, the `Classifier` part is fixed, and the `Encoder-Decoder` part is trained to fit the classifier, along with continuing to minimize the autoencoding loss.

(**Note:** When using your own dataset, make sure to set `max_decoding_length_train` and `max_decoding_length_infer` in [config.py](https://github.com/asyml/texar/blob/master/examples/text_style_transfer/config.py#L85-L86).)

Training log is printed as below:
```
gamma: 1.0, lambda_g: 0.0
Expand Down
2 changes: 1 addition & 1 deletion texar/tf/version.py
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@

_MAJOR = "0"
_MINOR = "2"
_REVISION = "3-unreleased"
_REVISION = "3"

VERSION_SHORT = "{0}.{1}".format(_MAJOR, _MINOR)
VERSION = "{0}.{1}.{2}".format(_MAJOR, _MINOR, _REVISION)

0 comments on commit 8dac88d

Please sign in to comment.