Fuxictr

+ [Fix] Fix regularization_loss() when feature_encoders exist ([135](https://github.com/reczoo/FuxiCTR/issues/135))

+ [Fix] Fix init_weights() for PretrainedEmbedding by modifying embedding_initializer ([126](https://github.com/reczoo/FuxiCTR/issues/126))
+ [Fix] Fix get_mask issue when num_heads > 1 ([130](https://github.com/reczoo/FuxiCTR/issues/130))
+ [Fix] Fix TransAct error when number of sequence features > 2 ([132](https://github.com/reczoo/FuxiCTR/issues/132))

+ [Fix] Fix get_inputs() bug ([115](https://github.com/reczoo/FuxiCTR/issues/115))

+ [Feature] Add WuKong model
+ [Fix] Fix OOV token update ([119](https://github.com/reczoo/FuxiCTR/issues/119))
+ [Fix] Speed up parquet dataset reading ([121](https://github.com/reczoo/FuxiCTR/issues/121))
+ [Fix] Fix add_loss() isue that does not work after renaming to compute_loss() ([122](https://github.com/reczoo/FuxiCTR/issues/122))
+ [Fix] Rename customized reset_parameters to init_weights ([123](https://github.com/reczoo/FuxiCTR/issues/123))

+ [Feature] Add EulerNet and DCNv3 models
+ [Feature] Add support to parquet as input, like csv format
+ [Fix] Add col_name as default args in feature_preprocess ([105](https://github.com/reczoo/FuxiCTR/issues/105))
+ [Fix] Fix fill_na when copy from a column ([117](https://github.com/reczoo/FuxiCTR/issues/117))

+ [Feature] Add TransAct model
+ [Feature] Add new feature type `embedding`, supporting [`meta`, `numeric`, `embedding`, `categorical`, `sequence`]
+ [Fix] Fix typo error in copy_from of version v2.3.1
+ [Fix] Fix issue in build_dataset for skipping rebuilding dataset
+ [Fix] Fix typo error in DIEN (AUGRUCell->AGRUCell)
+ [Fix] Fix typo error in feature_embedding of tf version ([94](https://github.com/reczoo/FuxiCTR/issues/94))