v0.9.3

Change Log

Major Features and Improvements

• Add EDCN model @Heyi007

Bug Fixes and Other Changes

• fix typo in position encoding layer @aceport
• fix savedmodel error in tf2.4 #498 @shenweichen

DisscussionGroup

• Github Discussions
• Wechat Discussions

公众号：浅梦学习笔记	微信：deepctrbot	学习小组 加入 主题集合

v0.9.2

Change Log

Major Features and Improvements

• Support Python 3.9 and 3.10 @ArtificialZeng
• Support cos and ln attention_type in transformer layer @shenweichen

Bug Fixes and Other Changes

• fix typo in transformer layer #490 #494 @darryyoung @BlackcOVER

v0.9.1

Change Log

Major Features and Improvements

• Improve compatibility with tensorflow 2.x @shenweichen

Bug Fixes and Other Changes

• fix running error in tensorflow version higher than 2.5 #432 #453 @15pengyi

DisscussionGroup

• Github Discussions
• Wechat Discussions

公众号：浅梦学习笔记	微信：deepctrbot	学习小组 加入 主题集合

v0.9.0

Change Log

Major Features and Improvements

• add multitask learning models : link @morningsky @shenweichen
  • SharedBottom
  • ESMM(Entire Space Multi-task Model)
  • MMOE(Multi-gate Mixture-of-Experts)
  • PLE(Progressive Layered Extraction)
• add example of running multitask model in deepctr link

Bug Fixes and Other Changes

fix truncated feature value error when padding string sequence in run_multivalue_movielens_hash.py #389 @JiaweiCui

API changes

• set dnn_hidden_units parameter to (256, 128, 64)
• • add new packagedeepctr.models.multitask
• files of sequence model are moved from deepctr.models to deepctr.models.sequence
  • deepctr.models.din to deepctr.models.sequence.din
  • deepctr.models.dien to deepctr.models.sequence.dien
  • deepctr.models.dsin to deepctr.models.sequence.dsin
  • deepctr.models.bst to deepctr.models.sequence.bst

v0.8.7

Change Log

Major Features and Improvements

• Support pre-defined key-value vocabulary in Hash Layer example @dengc367

Bug Fixes and Other Changes

• Make position encoding in transformer consistent with original paper #379 @dengc367 @chenxli

API changes

Add vocabulary_path parameter for deepctr.feature_column.SparseFeat api doc

v0.8.6

Change Log

Major Features and Improvements

• add IFM and DIFM model @zanshuxun
• add DeepFEFM model @pandeconscious
• improve the compatibility for tensorflow 2.5 @shenweichen

Bug Fixes and Other Changes

• add training param to activation layer #360 @shenweichen
• add PositionEncoding Layer #365 @shenweichen
• fix error when linear_feature_columns is empty in tensorflow 1.12 @shenweichen

v0.8.5

Change Log

Major Features and Improvements

• Add BST model doc
• support scaled_dot_product,additive attention type and different pooling mode in Transformer layer. doc

API changes

• Add attention_type and output_type parameter for Transformer layer.

v0.8.4

Change Log

Bug Fixes and Other Changes

• fix error in dcn-mix

v0.8.3

Change Log

Major Features and Improvements

• Add DCN V2 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 @zanshuxun
• Add transform_fn for DenseFeat doc @workingloong

API changes

• signature ofDenseFeat 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. #487
• simplify methods in interaction.py

v0.8.2

Change Log

Major Features and Improvements

Refactor DNN Layer.