Deepctr

Change Log
Major Features and Improvements
- Add `BST` model [doc](https://deepctr-doc.readthedocs.io/en/latest/Features.html#bst-behavior-sequence-transformer)
- support `scaled_dot_product`,`additive` attention type and different pooling mode in `Transformer` layer. [doc](https://deepctr-doc.readthedocs.io/en/latest/deepctr.layers.sequence.html#deepctr.layers.sequence.Transformer)

API changes
- Add `attention_type` and `output_type` parameter for `Transformer` layer.

Change Log

Bug Fixes and Other Changes

- fix error in dcn-mix

Change Log
Major Features and Improvements
- Add [DCN V2](https://deepctr-doc.readthedocs.io/en/latest/Features.html#dcn-mix-improved-deep-cross-network-with-mix-of-experts-and-matrix-kernel) Chinese Introducton [DCN V2：Google提出改进版DCN，用于大规模排序系统的特征交叉学习](https://mp.weixin.qq.com/s/0qidwbxyfTkODTw2DIiRWw) zanshuxun
- Add `transform_fn` for `DenseFeat` [doc](https://deepctr-doc.readthedocs.io/en/latest/Features.html#densefeat) workingloong


API changes
- signature of`DenseFeat` changed
- before `(name, dimension, dtype)`
- now `(name, dimension, dtype, transform_fn)`
- The output shape of `BilinearInteraction` Layer used in `FiBiNET` changed
- before:``(batch_size, 1, filed_size*(filed_size-1)/2*embedding_size)``
- now: ``(batch_size, filed_size*(filed_size-1)/2, embedding_size)``

Bug Fixes and Other Changes
- fix model save issues in tf 2.1.0, h5py 3.0.0. https://github.com/shenweichen/DeepCTR/issues/487
- simplify methods in `interaction.py`

Change Log
Major Features and Improvements
Refactor `DNN` Layer.

Change Log
Major Features and Improvements

- Improve the reproducibility of models



Bug Fixes and Other Changes
1. Fix `ValueError: The name "seq_length" is used 2 times in the model. All layer names should be unique.` when running `run_dien.py` in tf-2.x https://github.com/shenweichen/DeepCTR/issues/186

2. Fix `ImportError: cannot import name 'create_embedding_matrix` when running
`from deepctr.inputs import create_embedding_matrix`
https://github.com/shenweichen/DeepCTR/issues/252

3. Fix `ValueError: The initial value's shape ((27, 4)) is not compatible with the explicitly supplied shape argument ((27, 1)).` when running `run_classification_criteo.py` in tf-2.3.0 https://github.com/shenweichen/DeepCTR/issues/259

Change Log [中文版本更新说明](https://zhuanlan.zhihu.com/p/151009317)
Major Features and Improvements
- Support `Tensorflow Estimator` for large scale data and distributed training.
- [example: Estimator with TFRecord](https://deepctr-doc.readthedocs.io/en/latest/Examples.html#estimator-with-tfrecord-classification-criteo)
- [example: Estimator with Pandas DataFrame](https://deepctr-doc.readthedocs.io/en/latest/Examples.html#estimator-with-pandas-dataframe-classification-criteo)
- [Estimator Models API](https://deepctr-doc.readthedocs.io/en/latest/Estimators.html)
- Support different initializers for different embedding weights by settting `embeddings_initializer` of `SparseFeat`
- Support using pretrained weights to initialize embedding weights and frozen embedding weights. [example](https://deepctr-doc.readthedocs.io/en/latest/FAQ.html#how-to-use-pretrained-weights-to-initialize-embedding-weights-and-frozen-embedding-weights)
- Add new model `FwFM`
Bug Fixes and Other Changes

API changes
- The followings are moved from `deepctr.inputs` to `deepctr.feature_column`
- Class:`SparseFeat`,`VarLenSparseFeat`,`DenseFeat`
- Function:`get_feature_names`,`build_input_features`,`get_linear_logit`,`input_from_feature_columns`
- Variable:`DEFAULT_GROUP_NAME `
- `combined_dnn_input` is moved from `deepctr.inputs` to `deepctr.layers.utils`
- `SparseFeat` add `embeddings_initializer` and `trainable` arguments。