├── .github └── workflows │ ├── contributor.yml │ ├── jupyter-book.yml │ └── pypi.yml ├── .gitignore ├── CHANGELOG.md ├── CITATION ├── CONTRIBUTORS.md ├── LICENSE ├── README.md ├── data ├── README.md ├── tiny_csv │ ├── test_sample.csv │ ├── train_sample.csv │ ├── userid_emb_dim8.npz │ └── valid_sample.csv ├── tiny_mtl │ ├── test.csv │ └── train.csv ├── tiny_npz │ ├── feature_map.json │ ├── test.npz │ ├── train.npz │ └── valid.npz ├── tiny_parquet │ ├── feature_map.json │ ├── test.parquet │ ├── train.parquet │ └── valid.parquet ├── tiny_seq │ ├── feature_map.json │ ├── test.npz │ ├── train.npz │ └── valid.npz └── tiny_tfrecord │ ├── feature_map.json │ ├── test.tfrecord │ ├── train.tfrecord │ └── valid.tfrecord ├── demo ├── config │ ├── example1_config │ │ └── dataset_config.yaml │ ├── example2_config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── example3_config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── example4_config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── example5_config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── example6_config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ └── example7_config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml ├── example1_build_dataset_to_parquet.py ├── example2_DeepFM_with_parquet_input.py ├── example3_DeepFM_with_npz_input.py ├── example4_DeepFM_with_csv_input.py ├── example5_DeepFM_with_pretrained_emb.py ├── example6_DIN_with_sequence_feature.py └── example7_DeepFM_with_customized_preprocess.py ├── docs ├── _config.yml ├── _toc.yml ├── configurations.md ├── img │ ├── DCN.PNG │ ├── MMoE.jpg │ ├── PLE.png │ ├── ShareBottom.jpg │ ├── logo.png │ └── workflow.jpg ├── index.md ├── installation.md ├── logo.png ├── quickstart.md ├── tutorials │ ├── v1.0 │ │ ├── _config.yml │ │ ├── _toc.yml │ │ ├── how_to_make_configurations.ipynb │ │ ├── img │ │ │ └── logo.png │ │ ├── index.md │ │ ├── install_fuxictr.ipynb │ │ ├── run_model_with_config_file.ipynb │ │ ├── run_model_with_h5_input.ipynb │ │ ├── run_the_demo.ipynb │ │ └── tune_model_via_grid_search.ipynb │ ├── v1.1 │ │ ├── _config.yml │ │ ├── _toc.yml │ │ ├── how_to_make_configurations.ipynb │ │ ├── img │ │ │ └── logo.png │ │ ├── index.md │ │ ├── install_fuxictr.ipynb │ │ ├── run_model_with_config_file.ipynb │ │ ├── run_model_with_h5_input.ipynb │ │ ├── run_the_demo.ipynb │ │ └── tune_model_via_grid_search.ipynb │ ├── v1.2 │ │ ├── _config.yml │ │ ├── _toc.yml │ │ ├── how_to_make_configurations.ipynb │ │ ├── img │ │ │ └── logo.png │ │ ├── index.md │ │ ├── install_fuxictr.ipynb │ │ ├── run_model_with_config_file.ipynb │ │ ├── run_model_with_h5_input.ipynb │ │ ├── run_the_demo.ipynb │ │ └── tune_model_via_grid_search.ipynb │ └── v2.0 │ │ ├── _config.yml │ │ ├── _toc.yml │ │ ├── citations.bib │ │ ├── configurations.md │ │ ├── img │ │ └── logo.png │ │ ├── index.md │ │ ├── installation.md │ │ └── quickstart.md └── workflow.jpg ├── experiment ├── config │ └── DCN_tiny_parquet_tuner_config.yaml ├── enumerate_param_list.py ├── fuxictr_version.py ├── run_expid.py └── run_param_tuner.py ├── fuxictr ├── __init__.py ├── autotuner.py ├── datasets │ ├── __init__.py │ ├── avazu.py │ ├── criteo.py │ └── kkbox.py ├── features.py ├── metrics.py ├── preprocess │ ├── __init__.py │ ├── build_dataset.py │ ├── feature_processor.py │ ├── normalizer.py │ └── tokenizer.py ├── pytorch │ ├── __init__.py │ ├── dataloaders │ │ ├── __init__.py │ │ ├── npz_block_dataloader.py │ │ ├── npz_dataloader.py │ │ ├── parquet_block_dataloader.py │ │ ├── parquet_dataloader.py │ │ └── rank_dataloader.py │ ├── layers │ │ ├── __init__.py │ │ ├── activations.py │ │ ├── attentions │ │ │ ├── __init__.py │ │ │ ├── dot_product_attention.py │ │ │ ├── squeeze_excitation.py │ │ │ └── target_attention.py │ │ ├── blocks │ │ │ ├── __init__.py │ │ │ ├── factorization_machine.py │ │ │ ├── logistic_regression.py │ │ │ └── mlp_block.py │ │ ├── embeddings │ │ │ ├── __init__.py │ │ │ ├── feature_embedding.py │ │ │ └── pretrained_embedding.py │ │ ├── interactions │ │ │ ├── __init__.py │ │ │ ├── bilinear_interaction.py │ │ │ ├── compressed_interaction_net.py │ │ │ ├── cross_net.py │ │ │ ├── holographic_interaction.py │ │ │ ├── inner_product.py │ │ │ └── interaction_machine.py │ │ └── pooling.py │ ├── models │ │ ├── __init__.py │ │ ├── multitask_model.py │ │ └── rank_model.py │ └── torch_utils.py ├── tensorflow │ ├── __init__.py │ ├── dataloaders │ │ ├── __init__.py │ │ └── tf_dataloader.py │ ├── layers │ │ ├── __init__.py │ │ ├── blocks │ │ │ ├── __init__.py │ │ │ ├── factorization_machine.py │ │ │ ├── linear.py │ │ │ ├── logistic_regression.py │ │ │ └── mlp_block.py │ │ ├── embeddings │ │ │ ├── __init__.py │ │ │ └── feature_embedding.py │ │ ├── interactions │ │ │ ├── __init__.py │ │ │ ├── cross_net.py │ │ │ └── inner_product.py │ │ └── pooling.py │ ├── models │ │ ├── __init__.py │ │ └── rank_model.py │ └── tf_utils.py ├── utils.py └── version.py ├── model_zoo ├── AFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── AFM.py │ │ └── __init__.py ├── AFN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── AFN.py │ │ └── __init__.py ├── AOANet │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── AOANet.py │ │ └── __init__.py ├── APG │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── APG.py │ │ ├── APG_DCNv2.py │ │ ├── APG_DeepFM.py │ │ └── __init__.py ├── AutoInt │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── AutoInt.py │ │ └── __init__.py ├── BST │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── BST.py │ │ └── __init__.py ├── CCPM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── CCPM.py │ │ └── __init__.py ├── DCN │ ├── DCN_tf │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── DCN.py │ │ │ └── __init__.py │ └── DCN_torch │ │ ├── README.md │ │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ ├── DCN.py │ │ └── __init__.py ├── DCNv2 │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DCNv2.py │ │ └── __init__.py ├── DESTINE │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DESTINE.py │ │ └── __init__.py ├── DIEN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DIEN.py │ │ └── __init__.py ├── DIN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DIN.py │ │ └── __init__.py ├── DLRM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DLRM.py │ │ └── __init__.py ├── DMIN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DMIN.py │ │ └── __init__.py ├── DMR │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DMR.py │ │ └── __init__.py ├── DNN │ ├── DNN_tf │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── DNN.py │ │ │ └── __init__.py │ └── DNN_torch │ │ ├── README.md │ │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ ├── DNN.py │ │ └── __init__.py ├── DSSM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DSSM.py │ │ └── __init__.py ├── DeepCrossing │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DeepCrossing.py │ │ └── __init__.py ├── DeepFM │ ├── DeepFM_tf │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── DeepFM.py │ │ │ └── __init__.py │ └── DeepFM_torch │ │ ├── README.md │ │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ ├── DeepFM.py │ │ └── __init__.py ├── DeepIM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DeepIM.py │ │ └── __init__.py ├── EDCN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── EDCN.py │ │ └── __init__.py ├── EulerNet │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── readme.md │ ├── run_expid.py │ └── src │ │ ├── EulerNet.py │ │ └── __init__.py ├── FFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FFM.py │ │ ├── FFMv2.py │ │ └── __init__.py ├── FGCNN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FGCNN.py │ │ └── __init__.py ├── FLEN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FLEN.py │ │ └── __init__.py ├── FM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FM.py │ │ └── __init__.py ├── FiBiNET │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FiBiNET.py │ │ └── __init__.py ├── FiGNN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FiGNN.py │ │ └── __init__.py ├── FinalMLP │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── DualMLP.py │ │ ├── FinalMLP.py │ │ └── __init__.py ├── FinalNet │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FinalNet.py │ │ └── __init__.py ├── FmFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FmFM.py │ │ └── __init__.py ├── FwFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── FwFM.py │ │ └── __init__.py ├── GDCN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── GDCN.py │ │ └── __init__.py ├── HFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── HFM.py │ │ └── __init__.py ├── HOFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── HOFM.py │ │ └── __init__.py ├── InterHAt │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── InterHAt.py │ │ └── __init__.py ├── LR │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── LR.py │ │ └── __init__.py ├── LongCTR │ ├── DCNv2 │ │ ├── DCNv2.py │ │ └── __init__.py │ ├── DIEN │ │ ├── DIEN.py │ │ └── __init__.py │ ├── DIN │ │ ├── DIN.py │ │ └── __init__.py │ ├── ETA │ │ ├── ETA.py │ │ └── __init__.py │ ├── FinalMLP │ │ ├── FinalMLP.py │ │ └── __init__.py │ ├── MIRRN │ │ ├── MIRRN.py │ │ └── __init__.py │ ├── SDIM │ │ ├── SDIM.py │ │ └── __init__.py │ ├── SIM │ │ ├── SIM.py │ │ └── __init__.py │ ├── TWIN │ │ ├── TWIN.py │ │ └── __init__.py │ ├── TransAct │ │ ├── TransAct.py │ │ └── __init__.py │ ├── __init__.py │ ├── config │ │ ├── dcnv2_config │ │ │ └── model_config.yaml │ │ ├── dien_config │ │ │ └── model_config.yaml │ │ ├── din_config │ │ │ └── model_config.yaml │ │ ├── eta_config │ │ │ └── model_config.yaml │ │ ├── finalmlp_config │ │ │ └── model_config.yaml │ │ ├── mirrn_config │ │ │ └── model_config.yaml │ │ ├── sdim_config │ │ │ └── model_config.yaml │ │ ├── sim_config │ │ │ └── model_config.yaml │ │ ├── transact_config │ │ │ └── model_config.yaml │ │ └── twin_config │ │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── longctr_dataloader.py │ └── run_expid.py ├── LorentzFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── LorentzFM.py │ │ └── __init__.py ├── MaskNet │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── MaskNet.py │ │ └── __init__.py ├── NFM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── NFM.py │ │ └── __init__.py ├── ONN │ ├── ONN_tf │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── ONN.py │ │ │ └── __init__.py │ └── ONN_torch │ │ ├── README.md │ │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ ├── ONN.py │ │ ├── ONNv2.py │ │ └── __init__.py ├── PEPNet │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── PEPNet.py │ │ ├── PPNet.py │ │ └── __init__.py ├── PNN │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── PNN.py │ │ └── __init__.py ├── SAM │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── SAM.py │ │ └── __init__.py ├── TransAct │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── TransAct.py │ │ └── __init__.py ├── WideDeep │ ├── WideDeep_tf │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── WideDeep.py │ │ │ └── __init__.py │ └── WideDeep_torch │ │ ├── README.md │ │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ ├── WideDeep.py │ │ └── __init__.py ├── WuKong │ ├── README.md │ ├── config │ │ ├── dataset_config.yaml │ │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ │ ├── WuKong.py │ │ └── __init__.py ├── __init__.py ├── multitask │ ├── MMoE │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── MMoE.py │ │ │ └── __init__.py │ ├── PLE │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── PLE.py │ │ │ └── __init__.py │ ├── ShareBottom │ │ ├── README.md │ │ ├── config │ │ │ ├── dataset_config.yaml │ │ │ └── model_config.yaml │ │ ├── fuxictr_version.py │ │ ├── run_expid.py │ │ └── src │ │ │ ├── ShareBottom.py │ │ │ └── __init__.py │ └── __init__.py └── xDeepFM │ ├── README.md │ ├── config │ ├── dataset_config.yaml │ └── model_config.yaml │ ├── fuxictr_version.py │ ├── run_expid.py │ └── src │ ├── __init__.py │ └── xDeepFM.py ├── requirements.txt ├── setup.py └── tests ├── test_demo.sh ├── test_tensorflow.sh ├── test_torch.sh └── unit_tests └── test_torch_layers_bilinear_interaction.py /.github/workflows/contributor.yml: -------------------------------------------------------------------------------- 1 | name: Contributor List 2 | 3 | on: 4 | pull_request: 5 | branches: 6 | - main 7 | types: [closed] 8 | workflow_dispatch: 9 | 10 | jobs: 11 | contrib-readme-job: 12 | runs-on: ubuntu-latest 13 | steps: 14 | - name: Add contributor list 15 | uses: akhilmhdh/contributors-readme-action@master 16 | with: 17 | readme_path: "CONTRIBUTORS.md" 18 | image_size: 80 19 | commit_message: "Automatically update contributors" 20 | columns_per_row: 6 21 | env: 22 | GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} 23 | -------------------------------------------------------------------------------- /.github/workflows/jupyter-book.yml: -------------------------------------------------------------------------------- 1 | name: jupyter-book 2 | 3 | # Only run this when the master branch changes 4 | on: 5 | push: 6 | branches: 7 | - main 8 | 9 | # This job installs dependencies, build the book, and pushes it to `gh-pages` 10 | jobs: 11 | deploy-book: 12 | runs-on: ubuntu-latest 13 | steps: 14 | # Install dependencies 15 | - uses: actions/checkout@v3 16 | - uses: actions/setup-python@v3 17 | with: 18 | python-version: 3.7 19 | - run: pip install jupyter-book 20 | 21 | # # Build the book 22 | # - name: Build the book 23 | # run: | 24 | # jupyter-book build ./docs/tutorials/v2.0 25 | 26 | # # Push the book's HTML to github-pages 27 | # - name: Deploy github-pages 28 | # uses: peaceiris/actions-gh-pages@v3 29 | # with: 30 | # personal_token: ${{ secrets.PERSONAL_TOKEN }} 31 | # external_repository: fuxictr/fuxictr.github.io 32 | # publish_branch: main 33 | # publish_dir: ./docs/tutorials/v2.0/_build/html 34 | # exclude_assets: '_sources' 35 | # destination_dir: tutorials/v2.0 36 | -------------------------------------------------------------------------------- /.github/workflows/pypi.yml: -------------------------------------------------------------------------------- 1 | name: Release pypi wheels 2 | 3 | on: 4 | push: 5 | tags: 6 | - '*' 7 | workflow_dispatch: 8 | 9 | jobs: 10 | pypi-publish: 11 | name: Build package wheel and upload to pypi 12 | runs-on: ubuntu-latest 13 | # Specifying a GitHub environment is optional, but strongly encouraged 14 | environment: release 15 | permissions: 16 | # IMPORTANT: this permission is mandatory for trusted publishing 17 | id-token: write 18 | steps: 19 | - uses: actions/checkout@v4 20 | - name: Set up Python 21 | uses: actions/setup-python@v4 22 | with: 23 | python-version: 3.9 24 | - name: Build package wheel and source tarball 25 | run: | 26 | pip install wheel 27 | python setup.py sdist bdist_wheel 28 | - name: Publish package distributions to pypi 29 | uses: pypa/gh-action-pypi-publish@release/v1 30 | with: 31 | packages-dir: dist/ 32 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *.pyc 5 | *$py.class 6 | 7 | # C extensions 8 | *.so 9 | 10 | # Distribution / packaging 11 | .Python 12 | build/ 13 | develop-eggs/ 14 | dist/ 15 | downloads/ 16 | eggs/ 17 | .eggs/ 18 | lib/ 19 | lib64/ 20 | parts/ 21 | sdist/ 22 | var/ 23 | wheels/ 24 | *.egg-info/ 25 | .installed.cfg 26 | *.egg 27 | MANIFEST* 28 | 29 | # PyInstaller 30 | # Usually these files are written by a python script from a template 31 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 32 | *.manifest 33 | *.spec 34 | 35 | # Installer logs 36 | pip-log.txt 37 | pip-delete-this-directory.txt 38 | 39 | # Unit test / coverage reports 40 | htmlcov/ 41 | .tox/ 42 | .coverage 43 | .coverage.* 44 | .cache 45 | nosetests.xml 46 | coverage.xml 47 | *.cover 48 | .hypothesis/ 49 | .pytest_cache/ 50 | 51 | # Translations 52 | *.mo 53 | *.pot 54 | 55 | # Django stuff: 56 | *.log 57 | local_settings.py 58 | db.sqlite3 59 | 60 | # Flask stuff: 61 | instance/ 62 | .webassets-cache 63 | 64 | # Scrapy stuff: 65 | .scrapy 66 | 67 | # Sphinx documentation 68 | docs/_build/ 69 | 70 | # PyBuilder 71 | target/ 72 | 73 | # Jupyter Notebook 74 | .ipynb_checkpoints 75 | 76 | # pyenv 77 | .python-version 78 | 79 | # celery beat schedule file 80 | celerybeat-schedule 81 | 82 | # SageMath parsed files 83 | *.sage.py 84 | 85 | # Environments 86 | .env 87 | .venv 88 | env/ 89 | venv/ 90 | ENV/ 91 | env.bak/ 92 | venv.bak/ 93 | 94 | # Spyder project settings 95 | .spyderproject 96 | .spyproject 97 | 98 | # Rope project settings 99 | .ropeproject 100 | 101 | # mkdocs documentation 102 | /site 103 | 104 | # mypy 105 | .mypy_cache/ 106 | _build 107 | checkpoints 108 | *.csv 109 | -------------------------------------------------------------------------------- /CITATION: -------------------------------------------------------------------------------- 1 | @incollection{FuxiCTR, 2 | author = {Jieming Zhu and 3 | Jinyang Liu and 4 | Shuai Yang and 5 | Qi Zhang and 6 | Xiuqiang He}, 7 | title = {Open Benchmarking for Click-Through Rate Prediction}, 8 | booktitle = {The 30th {ACM} International Conference on Information 9 | and Knowledge Management (CIKM'21)}, 10 | pages = {2759--2769}, 11 | year = {2021} 12 | } 13 | 14 | @incollection{BARS, 15 | author = {Jieming Zhu and 16 | Quanyu Dai and 17 | Liangcai Su and 18 | Rong Ma and 19 | Jinyang Liu and 20 | Guohao Cai and 21 | Xi Xiao and 22 | Rui Zhang}, 23 | title = {BARS: Towards Open Benchmarking for Recommender Systems}, 24 | booktitle = {The 45th International ACM SIGIR Conference on Research 25 | and Development in Information Retrieval (SIGIR'22)}, 26 | year = {2022} 27 | } 28 | 29 | -------------------------------------------------------------------------------- /data/README.md: -------------------------------------------------------------------------------- 1 | ## Datasets 2 | 3 | A list of benchmark datasets for CTR prediction https://github.com/reczoo/Datasets 4 | -------------------------------------------------------------------------------- /data/tiny_csv/userid_emb_dim8.npz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_csv/userid_emb_dim8.npz -------------------------------------------------------------------------------- /data/tiny_npz/test.npz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_npz/test.npz -------------------------------------------------------------------------------- /data/tiny_npz/train.npz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_npz/train.npz -------------------------------------------------------------------------------- /data/tiny_npz/valid.npz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_npz/valid.npz -------------------------------------------------------------------------------- /data/tiny_parquet/test.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_parquet/test.parquet -------------------------------------------------------------------------------- /data/tiny_parquet/train.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_parquet/train.parquet -------------------------------------------------------------------------------- /data/tiny_parquet/valid.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_parquet/valid.parquet -------------------------------------------------------------------------------- /data/tiny_seq/test.npz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_seq/test.npz -------------------------------------------------------------------------------- /data/tiny_seq/train.npz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_seq/train.npz -------------------------------------------------------------------------------- /data/tiny_seq/valid.npz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_seq/valid.npz -------------------------------------------------------------------------------- /data/tiny_tfrecord/test.tfrecord: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_tfrecord/test.tfrecord -------------------------------------------------------------------------------- /data/tiny_tfrecord/train.tfrecord: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_tfrecord/train.tfrecord -------------------------------------------------------------------------------- /data/tiny_tfrecord/valid.tfrecord: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/data/tiny_tfrecord/valid.tfrecord -------------------------------------------------------------------------------- /demo/config/example1_config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for demo only 2 | tiny_example1: 3 | data_root: ../data/ 4 | data_format: csv 5 | train_data: ../data/tiny_csv/train_sample.csv 6 | valid_data: ../data/tiny_csv/valid_sample.csv 7 | test_data: ../data/tiny_csv/test_sample.csv 8 | min_categr_count: 1 9 | feature_cols: 10 | [{name: ["userid","adgroup_id","pid","cate_id","campaign_id","customer","brand","cms_segid", 11 | "cms_group_id","final_gender_code","age_level","pvalue_level","shopping_level","occupation"], 12 | active: True, dtype: str, type: categorical}] 13 | label_col: {name: clk, dtype: float} 14 | -------------------------------------------------------------------------------- /demo/config/example2_config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for demo only 2 | tiny_parquet: 3 | data_root: ../data/ 4 | data_format: parquet 5 | train_data: ../data/tiny_parquet/train.parquet 6 | valid_data: ../data/tiny_parquet/valid.parquet 7 | test_data: ../data/tiny_parquet/test.parquet 8 | -------------------------------------------------------------------------------- /demo/config/example2_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepFM_test_parquet: 16 | model: DeepFM 17 | dataset_id: tiny_parquet 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | hidden_units: [64, 32] 23 | hidden_activations: relu 24 | net_regularizer: 0 25 | embedding_regularizer: 1.e-8 26 | learning_rate: 1.e-3 27 | batch_norm: False 28 | net_dropout: 0 29 | batch_size: 128 30 | embedding_dim: 4 31 | epochs: 1 32 | shuffle: True 33 | seed: 2023 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | -------------------------------------------------------------------------------- /demo/config/example3_config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for demo only 2 | tiny_npz: 3 | data_root: ../data/ 4 | data_format: npz 5 | train_data: ../data/tiny_npz/train.npz 6 | valid_data: ../data/tiny_npz/valid.npz 7 | test_data: ../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /demo/config/example3_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepFM_test_npz: 16 | model: DeepFM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | hidden_units: [64, 32] 23 | hidden_activations: relu 24 | net_regularizer: 0 25 | embedding_regularizer: 1.e-8 26 | learning_rate: 1.e-3 27 | batch_norm: False 28 | net_dropout: 0 29 | batch_size: 128 30 | embedding_dim: 4 31 | epochs: 1 32 | shuffle: True 33 | seed: 2023 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | -------------------------------------------------------------------------------- /demo/config/example4_config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for demo only 2 | tiny_example4: 3 | data_root: ../data/ 4 | data_format: csv 5 | train_data: ../data/tiny_csv/train_sample.csv 6 | valid_data: ../data/tiny_csv/valid_sample.csv 7 | test_data: ../data/tiny_csv/test_sample.csv 8 | min_categr_count: 1 9 | feature_cols: 10 | [{name: ["userid","adgroup_id","pid","cate_id","campaign_id","customer","brand","cms_segid", 11 | "cms_group_id","final_gender_code","age_level","pvalue_level","shopping_level","occupation"], 12 | active: True, dtype: str, type: categorical}] 13 | label_col: {name: clk, dtype: float} 14 | -------------------------------------------------------------------------------- /demo/config/example4_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepFM_test_csv: 16 | model: DeepFM 17 | dataset_id: tiny_example4 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | hidden_units: [64, 32] 23 | hidden_activations: relu 24 | net_regularizer: 0 25 | embedding_regularizer: 1.e-8 26 | learning_rate: 1.e-3 27 | batch_norm: False 28 | net_dropout: 0 29 | batch_size: 128 30 | embedding_dim: 4 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | -------------------------------------------------------------------------------- /demo/config/example5_config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for demo only 2 | tiny_example5: 3 | data_root: ../data/ 4 | data_format: csv 5 | train_data: ../data/tiny_csv/train_sample.csv 6 | valid_data: ../data/tiny_csv/valid_sample.csv 7 | test_data: ../data/tiny_csv/test_sample.csv 8 | min_categr_count: 1 9 | feature_cols: 10 | [{name: "userid", active: True, dtype: str, type: categorical, pretrained_emb: "../data/tiny_csv/userid_emb_dim8.npz", 11 | embedding_dim: 8, freeze_emb: True}, 12 | {name: ["adgroup_id","pid","cate_id","campaign_id","customer","brand","cms_segid", 13 | "cms_group_id","final_gender_code","age_level","pvalue_level","shopping_level","occupation"], 14 | active: True, dtype: str, type: categorical}] 15 | label_col: {name: clk, dtype: float} 16 | -------------------------------------------------------------------------------- /demo/config/example5_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepFM_test_pretrain: 16 | model: DeepFM 17 | dataset_id: tiny_example5 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | hidden_units: [64, 32] 23 | hidden_activations: relu 24 | net_regularizer: 0 25 | embedding_regularizer: 1.e-8 26 | learning_rate: 1.e-3 27 | batch_norm: False 28 | net_dropout: 0 29 | batch_size: 128 30 | embedding_dim: 8 31 | epochs: 1 32 | shuffle: True 33 | seed: 2023 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | -------------------------------------------------------------------------------- /demo/config/example6_config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for demo only 2 | tiny_seq: 3 | data_root: ../data/ 4 | data_format: npz 5 | train_data: ../data/tiny_seq/train.npz 6 | valid_data: ../data/tiny_seq/valid.npz 7 | test_data: ../data/tiny_seq/test.npz 8 | -------------------------------------------------------------------------------- /demo/config/example6_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DIN_test: 16 | model: DIN 17 | dataset_id: tiny_seq 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | attention_hidden_units: [64] 30 | attention_hidden_activations: "Dice" 31 | attention_output_activation: null 32 | attention_dropout: 0 33 | din_target_field: adgroup_id 34 | din_sequence_field: click_sequence 35 | feature_specs: [{name: click_sequence, feature_encoder: null}] 36 | net_dropout: 0 37 | batch_norm: False 38 | epochs: 1 39 | shuffle: True 40 | seed: 2019 41 | monitor: 'AUC' 42 | monitor_mode: 'max' 43 | -------------------------------------------------------------------------------- /demo/config/example7_config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for demo only 2 | tiny_example7: 3 | data_root: ../data/ 4 | data_format: csv 5 | train_data: ../data/tiny_csv/train_sample.csv 6 | valid_data: ../data/tiny_csv/valid_sample.csv 7 | test_data: ../data/tiny_csv/test_sample.csv 8 | min_categr_count: 1 9 | feature_cols: 10 | [{name: ["userid","adgroup_id","pid","cate_id","campaign_id","customer","brand","cms_segid", 11 | "cms_group_id","final_gender_code","age_level","pvalue_level","shopping_level","occupation"], 12 | active: True, dtype: str, type: categorical}, 13 | {name: "weekday", active: True, dtype: str, type: categorical, preprocess: convert_weekday}, 14 | {name: "hour", active: True, dtype: str, type: categorical, preprocess: convert_hour}] 15 | label_col: {name: clk, dtype: float} 16 | -------------------------------------------------------------------------------- /demo/config/example7_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepFM_test_csv: 16 | model: DeepFM 17 | dataset_id: tiny_example7 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | hidden_units: [64, 32] 23 | hidden_activations: relu 24 | net_regularizer: 0 25 | embedding_regularizer: 1.e-8 26 | learning_rate: 1.e-3 27 | batch_norm: False 28 | net_dropout: 0 29 | batch_size: 128 30 | embedding_dim: 4 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | -------------------------------------------------------------------------------- /docs/_toc.yml: -------------------------------------------------------------------------------- 1 | # Table of contents 2 | # Learn more at https://jupyterbook.org/customize/toc.html 3 | 4 | format: jb-book 5 | root: index 6 | parts: 7 | - caption: TUTORIALS 8 | chapters: 9 | - file: quickstart.md 10 | - file: installation.md 11 | - file: configurations.md 12 | 13 | - caption: REFERENCES 14 | chapters: 15 | - url: https://www.processon.com/view/link/63cfcfab4e30670eac4a81c7 16 | title: FuxiCTR APIs 17 | -------------------------------------------------------------------------------- /docs/img/DCN.PNG: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/img/DCN.PNG -------------------------------------------------------------------------------- /docs/img/MMoE.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/img/MMoE.jpg -------------------------------------------------------------------------------- /docs/img/PLE.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/img/PLE.png -------------------------------------------------------------------------------- /docs/img/ShareBottom.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/img/ShareBottom.jpg -------------------------------------------------------------------------------- /docs/img/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/img/logo.png -------------------------------------------------------------------------------- /docs/img/workflow.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/img/workflow.jpg -------------------------------------------------------------------------------- /docs/index.md: -------------------------------------------------------------------------------- 1 | # FuxiCTR Tutorials 2 | 3 | ```{tableofcontents} 4 | ``` 5 | 6 | ```{note} 7 | Tutorials for FuxiCTR v2 are under construction. 8 | ``` 9 | 10 | ## Previous Versions 11 | 12 | + [Tutorials v1.2](https://fuxictr.github.io/tutorials/v1.2) 13 | + [Tutorials v1.1](https://fuxictr.github.io/tutorials/v1.1) 14 | + [Tutorials v1.0](https://fuxictr.github.io/tutorials/v1.0) 15 | 16 | -------------------------------------------------------------------------------- /docs/installation.md: -------------------------------------------------------------------------------- 1 | # Install FuxiCTR 2 | 3 | ```{note} 4 | Tutorials for FuxiCTR v2 only. 5 | ``` 6 | 7 | FuxiCTR v2 has the following requirements. 8 | 9 | + pytorch 1.10+ (required only for torch models) 10 | + tensorflow 2.1+ (required only for tf models) 11 | + python 3.6+ 12 | + pyyaml 5.1+ 13 | + scikit-learn 14 | + pandas 15 | + numpy 16 | + h5py 17 | + tqdm 18 | 19 | We recommend to install the above enviornment with `python 3.7` through Anaconda using [Anaconda3-2020.02-Linux-x86_64.sh](https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/Anaconda3-2020.02-Linux-x86_64.sh). 20 | 21 | For pytorch, you can download the appropriate whl file according to your CUDA version from https://download.pytorch.org/whl/torch_stable.html, and install offline. For eaxmple: 22 | 23 | ``` 24 | pip install torch-1.11.0%2Bcu102-cp37-cp37m-linux_x86_64.whl 25 | ``` 26 | 27 | There are two ways to install FuxiCTR v2: 28 | 29 | **Solution 1**: pip install 30 | 31 | ``` 32 | pip install fuxictr==2.0.0 33 | ``` 34 | 35 | ```{note} 36 | All the dependent packages need to be installed accordingly. 37 | ``` 38 | 39 | **Solution 2**: git clone or download the zip file: https://github.com/xue-pai/FuxiCTR/tags 40 | 41 | If you download the source code, you need to add the fuxictr folder to the system path in your code. 42 | 43 | ```python 44 | import sys 45 | sys.path.append('./YOUR_PATH_TO_FuxiCTR') 46 | ``` 47 | 48 | Check if fuxictr has been installed successfully. 49 | 50 | ```python 51 | import fuxictr 52 | print(fuxictr.__version__) 53 | ``` 54 | -------------------------------------------------------------------------------- /docs/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/logo.png -------------------------------------------------------------------------------- /docs/quickstart.md: -------------------------------------------------------------------------------- 1 | # Quick Start 2 | 3 | FuxiCTR supports two types of usages as follows. 4 | 5 | 1. **To run an existing model**: Users can easily run each model in the model zoo following the commands below, which is a demo for running DCN. In addition, users can modify the dataset config and model config files to run on their own datasets or with new hyper-parameters. More details can be found in the [readme file](https://github.com/xue-pai/FuxiCTR/blob/v2.0.0/model_zoo/DCN/DCN_torch/README.md). 6 | ``` 7 | cd model_zoo/DCN/DCN_torch 8 | python run_expid.py --expid DCN_test --gpu 0 9 | ``` 10 | 11 | 2. **To implement a new model**: The FuxiCTR code structure is modularized, so that every part can be overwritten by users according to their needs. As the workflow shown in the following figure, the orange parts comprise the minimal user code to implement a new customized model. In case that data preprocessing or data loader is not directly applicable, one can overwrite a new one through the [core APIs](https://www.processon.com/view/link/63cfcfab4e30670eac4a81c7). Some examples can also be found in the model zoo. 12 | 13 |
14 | FuxiCTR Workflow 15 |
16 | -------------------------------------------------------------------------------- /docs/tutorials/v1.0/_toc.yml: -------------------------------------------------------------------------------- 1 | format: jb-book 2 | root: index 3 | parts: 4 | - caption: Tutorials 5 | chapters: 6 | - file: install_fuxictr.ipynb 7 | - file: run_the_demo.ipynb 8 | - file: run_model_with_config_file.ipynb 9 | - file: run_model_with_h5_input.ipynb 10 | - file: how_to_make_configurations.ipynb 11 | - file: tune_model_via_grid_search.ipynb 12 | -------------------------------------------------------------------------------- /docs/tutorials/v1.0/img/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/tutorials/v1.0/img/logo.png -------------------------------------------------------------------------------- /docs/tutorials/v1.0/index.md: -------------------------------------------------------------------------------- 1 | # FuxiCTR v1.0 Tutorials 2 | 3 | ```{note} 4 | The tutorials are built for FuxiCTR v1.0. 5 | ``` 6 | 7 | ```{tableofcontents} 8 | ``` 9 | 10 | -------------------------------------------------------------------------------- /docs/tutorials/v1.1/_toc.yml: -------------------------------------------------------------------------------- 1 | format: jb-book 2 | root: index 3 | parts: 4 | - caption: Tutorials 5 | chapters: 6 | - file: install_fuxictr.ipynb 7 | - file: run_the_demo.ipynb 8 | - file: run_model_with_config_file.ipynb 9 | - url: https://github.com/xue-pai/FuxiCTR/tree/v1.1.0/demo/preprocess_h5_demo.py 10 | title: Preprocess raw csv data to h5 data 11 | - file: run_model_with_h5_input.ipynb 12 | - file: how_to_make_configurations.ipynb 13 | - file: tune_model_via_grid_search.ipynb 14 | - url: https://github.com/xue-pai/FuxiCTR/tree/v1.1.0/demo/DeepFM_with_sequence_feature.py 15 | title: Run a model with sequence features 16 | - url: https://github.com/xue-pai/FuxiCTR/tree/v1.1.0/demo/DeepFM_with_pretrained_emb.py 17 | title: Run a model with pretrained embeddings 18 | 19 | - caption: 中文教程 20 | chapters: 21 | - url: https://zhuanlan.zhihu.com/p/437373335 22 | title: CTR模型的高效、高性能实现（一） 23 | - url: https://zhuanlan.zhihu.com/p/453385054 24 | title: CTR模型的高效、高性能实现（二） -------------------------------------------------------------------------------- /docs/tutorials/v1.1/img/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/tutorials/v1.1/img/logo.png -------------------------------------------------------------------------------- /docs/tutorials/v1.1/index.md: -------------------------------------------------------------------------------- 1 | # FuxiCTR v1.1 Tutorials 2 | 3 | ```{note} 4 | The tutorials are built for FuxiCTR v1.1. 5 | ``` 6 | 7 | ```{tableofcontents} 8 | ``` 9 | 10 | -------------------------------------------------------------------------------- /docs/tutorials/v1.2/_toc.yml: -------------------------------------------------------------------------------- 1 | format: jb-book 2 | root: index 3 | parts: 4 | - caption: Tutorials 5 | chapters: 6 | - file: install_fuxictr.ipynb 7 | - file: run_the_demo.ipynb 8 | - file: run_model_with_config_file.ipynb 9 | - url: https://github.com/xue-pai/FuxiCTR/tree/v1.2.0/demo/preprocess_h5_demo.py 10 | title: Preprocess raw csv data to h5 data 11 | - file: run_model_with_h5_input.ipynb 12 | - file: how_to_make_configurations.ipynb 13 | - file: tune_model_via_grid_search.ipynb 14 | - url: https://github.com/xue-pai/FuxiCTR/tree/v1.2.0/demo/DeepFM_with_sequence_feature.py 15 | title: Run a model with sequence features 16 | - url: https://github.com/xue-pai/FuxiCTR/tree/v1.2.0/demo/DeepFM_with_pretrained_emb.py 17 | title: Run a model with pretrained embeddings 18 | 19 | - caption: 中文教程 20 | chapters: 21 | - url: https://zhuanlan.zhihu.com/p/437373335 22 | title: CTR模型的高效、高性能实现（一） 23 | - url: https://zhuanlan.zhihu.com/p/453385054 24 | title: CTR模型的高效、高性能实现（二） -------------------------------------------------------------------------------- /docs/tutorials/v1.2/img/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/tutorials/v1.2/img/logo.png -------------------------------------------------------------------------------- /docs/tutorials/v1.2/index.md: -------------------------------------------------------------------------------- 1 | # FuxiCTR v1.2 Tutorials 2 | 3 | ```{note} 4 | The tutorials are built for FuxiCTR v1.2. 5 | ``` 6 | 7 | ```{tableofcontents} 8 | ``` 9 | -------------------------------------------------------------------------------- /docs/tutorials/v2.0/_toc.yml: -------------------------------------------------------------------------------- 1 | # Table of contents 2 | # Learn more at https://jupyterbook.org/customize/toc.html 3 | 4 | format: jb-book 5 | root: index 6 | parts: 7 | - caption: Tutorials 8 | chapters: 9 | - file: quickstart.md 10 | - file: installation.md 11 | - file: configurations.md 12 | 13 | - caption: API 14 | chapters: 15 | - url: https://www.processon.com/view/link/63cfcfab4e30670eac4a81c7 16 | title: FuxiCTR core APIs 17 | 18 | # - caption: 中文教程 19 | # chapters: 20 | # - file: todo 21 | 22 | -------------------------------------------------------------------------------- /docs/tutorials/v2.0/img/logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/tutorials/v2.0/img/logo.png -------------------------------------------------------------------------------- /docs/tutorials/v2.0/index.md: -------------------------------------------------------------------------------- 1 | # FuxiCTR Tutorials 2 | 3 | ```{tableofcontents} 4 | ``` 5 | 6 | ```{note} 7 | Tutorials for FuxiCTR v2 are under construction. 8 | ``` 9 | 10 | ## Previous Versions 11 | 12 | + [Tutorials v1.2](https://fuxictr.github.io/tutorials/v1.2) 13 | + [Tutorials v1.1](https://fuxictr.github.io/tutorials/v1.1) 14 | + [Tutorials v1.0](https://fuxictr.github.io/tutorials/v1.0) 15 | 16 | -------------------------------------------------------------------------------- /docs/tutorials/v2.0/installation.md: -------------------------------------------------------------------------------- 1 | # Install FuxiCTR 2 | 3 | ```{note} 4 | Tutorials for FuxiCTR v2 only. 5 | ``` 6 | 7 | FuxiCTR v2 has the following requirements. 8 | 9 | + pytorch 1.10+ (required only for torch models) 10 | + tensorflow 2.1+ (required only for tf models) 11 | + python 3.6+ 12 | + pyyaml 5.1+ 13 | + scikit-learn 14 | + pandas 15 | + numpy 16 | + h5py 17 | + tqdm 18 | 19 | We recommend to install the above enviornment with `python 3.7` through Anaconda using [Anaconda3-2020.02-Linux-x86_64.sh](https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/Anaconda3-2020.02-Linux-x86_64.sh). 20 | 21 | For pytorch, you can download the appropriate whl file according to your CUDA version from https://download.pytorch.org/whl/torch_stable.html, and install offline. For eaxmple: 22 | 23 | ``` 24 | pip install torch-1.11.0%2Bcu102-cp37-cp37m-linux_x86_64.whl 25 | ``` 26 | 27 | There are two ways to install FuxiCTR v2: 28 | 29 | **Solution 1**: pip install 30 | 31 | ``` 32 | pip install fuxictr==2.0.0 33 | ``` 34 | 35 | ```{note} 36 | All the dependent packages need to be installed accordingly. 37 | ``` 38 | 39 | **Solution 2**: git clone or download the zip file: https://github.com/xue-pai/FuxiCTR/tags 40 | 41 | If you download the source code, you need to add the fuxictr folder to the system path in your code. 42 | 43 | ```python 44 | import sys 45 | sys.path.append('./YOUR_PATH_TO_FuxiCTR') 46 | ``` 47 | 48 | Check if fuxictr has been installed successfully. 49 | 50 | ```python 51 | import fuxictr 52 | print(fuxictr.__version__) 53 | ``` 54 | -------------------------------------------------------------------------------- /docs/tutorials/v2.0/quickstart.md: -------------------------------------------------------------------------------- 1 | # Quick Start 2 | 3 | FuxiCTR supports two types of usages as follows. 4 | 5 | 1. **To run an existing model**: Users can easily run each model in the model zoo following the commands below, which is a demo for running DCN. In addition, users can modify the dataset config and model config files to run on their own datasets or with new hyper-parameters. More details can be found in the [readme file](https://github.com/xue-pai/FuxiCTR/blob/v2.0.0/model_zoo/DCN/DCN_torch/README.md). 6 | ``` 7 | cd model_zoo/DCN/DCN_torch 8 | python run_expid.py --expid DCN_test --gpu 0 9 | ``` 10 | 11 | 2. **To implement a new model**: The FuxiCTR code structure is modularized, so that every part can be overwritten by users according to their needs. As the workflow shown in the following figure, the orange parts comprise the minimal user code to implement a new customized model. In case that data preprocessing or data loader is not directly applicable, one can overwrite a new one through the [core APIs](https://www.processon.com/view/link/63cfcfab4e30670eac4a81c7). Some examples can also be found in the model zoo. 12 | 13 |
14 | FuxiCTR Workflow 15 |
16 | 17 | 18 | 19 | -------------------------------------------------------------------------------- /docs/workflow.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/docs/workflow.jpg -------------------------------------------------------------------------------- /experiment/config/DCN_tiny_parquet_tuner_config.yaml: -------------------------------------------------------------------------------- 1 | base_config: ../model_zoo/DCN/DCN_torch/config/ 2 | base_expid: DCN_default 3 | dataset_id: tiny_parquet 4 | 5 | tiny_parquet: 6 | data_root: ../data/ 7 | data_format: npz 8 | train_data: ../data/tiny_parquet/train.parquet 9 | valid_data: ../data/tiny_parquet/valid.parquet 10 | test_data: ../data/tiny_parquet/test.parquet 11 | 12 | tuner_space: 13 | model_root: './checkpoints/' 14 | embedding_dim: 16 15 | dnn_hidden_units: [[1024, 512, 256]] 16 | num_cross_layers: [5, 4] 17 | embedding_regularizer: [1.e-4, 5.e-4, 1.e-3, 5.e-3] 18 | net_dropout: [0.2, 0.1, 0] 19 | batch_norm: False 20 | learning_rate: 5.e-4 21 | batch_size: 1024 22 | seed: 20222023 23 | group_id: user_id 24 | metrics: [[gAUC, AUC, logloss]] 25 | monitor: {"gAUC": 1, "AUC": 1} 26 | -------------------------------------------------------------------------------- /experiment/enumerate_param_list.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # 4 | # Licensed under the Apache License, Version 2.0 (the "License"); 5 | # you may not use this file except in compliance with the License. 6 | # You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | # ========================================================================= 16 | 17 | from datetime import datetime 18 | import gc 19 | import pandas as pd 20 | import argparse 21 | import fuxictr_version 22 | from fuxictr import autotuner 23 | 24 | if __name__ == '__main__': 25 | parser = argparse.ArgumentParser() 26 | parser.add_argument('--config', type=str, required=True, 27 | default='../config/tuner_config.yaml', 28 | help='The config file for para tuning.') 29 | parser.add_argument('--exclude', type=str, default='', 30 | help='The experiment_result.csv file to exclude finished expid.') 31 | args = vars(parser.parse_args()) 32 | exclude_expid = [] 33 | if args['exclude'] != '': 34 | result_df = pd.read_csv(args['exclude'], header=None) 35 | expid_df = result_df.iloc[:, 2].map(lambda x: x.replace('[exp_id] ', '')) 36 | exclude_expid = expid_df.tolist() 37 | # generate parameter space combinations 38 | config_dir = autotuner.enumerate_params(args['config'], exclude_expid=exclude_expid) 39 | 40 | -------------------------------------------------------------------------------- /experiment/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import sys 3 | sys.path.append("../") 4 | import fuxictr 5 | assert fuxictr.__version__ >= "2.3.7" 6 | -------------------------------------------------------------------------------- /experiment/run_param_tuner.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2025. FuxiCTR Authors. All rights reserved. 3 | # 4 | # Licensed under the Apache License, Version 2.0 (the "License"); 5 | # you may not use this file except in compliance with the License. 6 | # You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | # ========================================================================= 16 | 17 | 18 | import argparse 19 | import os 20 | import fuxictr_version 21 | from fuxictr import autotuner 22 | 23 | if __name__ == '__main__': 24 | parser = argparse.ArgumentParser() 25 | parser.add_argument('--config', type=str, default='../config/tuner_config.yaml', 26 | help='The config directory or file path for para tuning.') 27 | parser.add_argument('--tag', type=str, default=None, 28 | help='Use the tag to determine which expid to run, e.g. 001 for the first expid.') 29 | parser.add_argument('--gpu', nargs='+', default=[-1], help='The list of gpu devices, -1 for cpu.') 30 | args = vars(parser.parse_args()) 31 | gpu_list = args['gpu'] 32 | expid_tag = args['tag'] 33 | 34 | # generate parameter space combinations 35 | if os.path.isdir(args['config']): 36 | config_dir = args["config"] 37 | else: 38 | config_dir = autotuner.enumerate_params(args['config']) 39 | autotuner.grid_search(config_dir, gpu_list, expid_tag) 40 | -------------------------------------------------------------------------------- /fuxictr/__init__.py: -------------------------------------------------------------------------------- 1 | from .version import __version__ -------------------------------------------------------------------------------- /fuxictr/datasets/__init__.py: -------------------------------------------------------------------------------- 1 | from . import criteo 2 | from . import avazu 3 | from . import kkbox 4 | -------------------------------------------------------------------------------- /fuxictr/datasets/avazu.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | from fuxictr.preprocess import FeatureProcessor 20 | from datetime import datetime, date 21 | import polars as pl 22 | 23 | 24 | class CustomizedFeatureProcessor(FeatureProcessor): 25 | def convert_weekday(self, col_name=None): 26 | def _convert_weekday(timestamp): 27 | dt = date(int('20' + timestamp[0:2]), int(timestamp[2:4]), int(timestamp[4:6])) 28 | return int(dt.strftime('%w')) 29 | return pl.col("hour").apply(_convert_weekday) 30 | 31 | def convert_weekend(self, col_name=None): 32 | def _convert_weekend(timestamp): 33 | dt = date(int('20' + timestamp[0:2]), int(timestamp[2:4]), int(timestamp[4:6])) 34 | return 1 if dt.strftime('%w') in ['6', '0'] else 0 35 | return pl.col("hour").apply(_convert_weekend) 36 | 37 | def convert_hour(self, col_name=None): 38 | return pl.col("hour").apply(lambda x: int(x[6:8])) 39 | -------------------------------------------------------------------------------- /fuxictr/datasets/criteo.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | import numpy as np 20 | from fuxictr.preprocess import FeatureProcessor 21 | import polars as pl 22 | 23 | 24 | class CustomizedFeatureProcessor(FeatureProcessor): 25 | def convert_to_bucket(self, col_name): 26 | def _convert_to_bucket(value): 27 | if value > 2: 28 | value = int(np.floor(np.log(value) ** 2)) 29 | else: 30 | value = int(value) 31 | return value 32 | return pl.col(col_name).apply(_convert_to_bucket).cast(pl.Int32) 33 | -------------------------------------------------------------------------------- /fuxictr/preprocess/__init__.py: -------------------------------------------------------------------------------- 1 | from .build_dataset import * 2 | from .feature_processor import * -------------------------------------------------------------------------------- /fuxictr/preprocess/normalizer.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | import numpy as np 19 | import sklearn.preprocessing as sklearn_preprocess 20 | 21 | 22 | class Normalizer(object): 23 | def __init__(self, normalizer): 24 | if not callable(normalizer): 25 | self.callable = False 26 | if normalizer in ['StandardScaler', 'MinMaxScaler']: 27 | self.normalizer = getattr(sklearn_preprocess, normalizer)() 28 | else: 29 | raise NotImplementedError('normalizer={}'.format(normalizer)) 30 | else: 31 | # normalizer is a method 32 | self.normalizer = normalizer 33 | self.callable = True 34 | 35 | def fit(self, X): 36 | if not self.callable: 37 | self.normalizer.fit(X.reshape(-1, 1)) 38 | 39 | def transform(self, X): 40 | if self.callable: 41 | return self.normalizer(X) 42 | else: 43 | return self.normalizer.transform(X.reshape(-1, 1)).flatten() 44 | -------------------------------------------------------------------------------- /fuxictr/pytorch/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/fuxictr/pytorch/__init__.py -------------------------------------------------------------------------------- /fuxictr/pytorch/dataloaders/__init__.py: -------------------------------------------------------------------------------- 1 | from .rank_dataloader import RankDataLoader 2 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/__init__.py: -------------------------------------------------------------------------------- 1 | from .pooling import * 2 | from .embeddings import * 3 | from .activations import * 4 | from .blocks import * 5 | from .interactions import * 6 | from .attentions import * 7 | 8 | 9 | 10 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/activations.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | import torch 20 | from torch import nn 21 | import numpy as np 22 | 23 | 24 | class Dice(nn.Module): 25 | def __init__(self, input_dim, eps=1e-9): 26 | super(Dice, self).__init__() 27 | self.bn = nn.BatchNorm1d(input_dim, affine=False, eps=eps, momentum=0.01) 28 | self.alpha = nn.Parameter(torch.zeros(input_dim)) 29 | 30 | def forward(self, X): 31 | p = torch.sigmoid(self.bn(X)) 32 | output = p * X + self.alpha * (1 - p) * X 33 | return output 34 | 35 | 36 | class GELU(nn.Module): 37 | def __init__(self): 38 | super(GELU, self).__init__() 39 | 40 | def forward(self, x): 41 | return 0.5 * x * (1 + torch.tanh(np.sqrt(2 / np.pi) * ( x + 0.044715 * torch.pow(x, 3)))) 42 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/attentions/__init__.py: -------------------------------------------------------------------------------- 1 | from .dot_product_attention import * 2 | from .squeeze_excitation import * 3 | from .target_attention import * 4 | 5 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/attentions/squeeze_excitation.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | import torch 20 | from torch import nn 21 | 22 | 23 | class SqueezeExcitation(nn.Module): 24 | def __init__(self, num_fields, reduction_ratio=3, excitation_activation="ReLU"): 25 | super(SqueezeExcitation, self).__init__() 26 | reduced_size = max(1, int(num_fields / reduction_ratio)) 27 | excitation = [nn.Linear(num_fields, reduced_size, bias=False), 28 | nn.ReLU(), 29 | nn.Linear(reduced_size, num_fields, bias=False)] 30 | if excitation_activation.lower() == "relu": 31 | excitation.append(nn.ReLU()) 32 | elif excitation_activation.lower() == "sigmoid": 33 | excitation.append(nn.Sigmoid()) 34 | else: 35 | raise NotImplementedError 36 | self.excitation = nn.Sequential(*excitation) 37 | 38 | def forward(self, feature_emb): 39 | Z = torch.mean(feature_emb, dim=-1, out=None) 40 | A = self.excitation(Z) 41 | V = feature_emb * A.unsqueeze(-1) 42 | return V 43 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/blocks/__init__.py: -------------------------------------------------------------------------------- 1 | from .logistic_regression import * 2 | from .factorization_machine import * 3 | from .mlp_block import * 4 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/blocks/factorization_machine.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | import torch 20 | from torch import nn 21 | from .logistic_regression import LogisticRegression 22 | from ..interactions import InnerProductInteraction 23 | 24 | 25 | class FactorizationMachine(nn.Module): 26 | def __init__(self, feature_map): 27 | super(FactorizationMachine, self).__init__() 28 | self.fm_layer = InnerProductInteraction(feature_map.num_fields, output="product_sum") 29 | self.lr_layer = LogisticRegression(feature_map, use_bias=True) 30 | 31 | def forward(self, X, feature_emb): 32 | lr_out = self.lr_layer(X) 33 | fm_out = self.fm_layer(feature_emb) 34 | output = fm_out + lr_out 35 | return output 36 | 37 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/blocks/logistic_regression.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | import torch 20 | from torch import nn 21 | from fuxictr.pytorch.layers import FeatureEmbedding 22 | 23 | 24 | class LogisticRegression(nn.Module): 25 | def __init__(self, feature_map, use_bias=True): 26 | super(LogisticRegression, self).__init__() 27 | self.bias = nn.Parameter(torch.zeros(1), requires_grad=True) if use_bias else None 28 | # A trick for quick one-hot encoding in LR 29 | self.embedding_layer = FeatureEmbedding(feature_map, 1, use_pretrain=False, use_sharing=False) 30 | 31 | def forward(self, X): 32 | embed_weights = self.embedding_layer(X) 33 | output = embed_weights.sum(dim=1) 34 | if self.bias is not None: 35 | output += self.bias 36 | return output 37 | 38 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/embeddings/__init__.py: -------------------------------------------------------------------------------- 1 | from .feature_embedding import * 2 | from .pretrained_embedding import * 3 | -------------------------------------------------------------------------------- /fuxictr/pytorch/layers/interactions/__init__.py: -------------------------------------------------------------------------------- 1 | from .inner_product import * 2 | from .holographic_interaction import * 3 | from .cross_net import * 4 | from .compressed_interaction_net import * 5 | from .bilinear_interaction import * 6 | from .inner_product import * 7 | from .interaction_machine import * 8 | 9 | -------------------------------------------------------------------------------- /fuxictr/pytorch/models/__init__.py: -------------------------------------------------------------------------------- 1 | from .rank_model import BaseModel 2 | from .multitask_model import MultiTaskModel -------------------------------------------------------------------------------- /fuxictr/tensorflow/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/fuxictr/tensorflow/__init__.py -------------------------------------------------------------------------------- /fuxictr/tensorflow/dataloaders/__init__.py: -------------------------------------------------------------------------------- 1 | from .tf_dataloader import TFRecordDataLoader -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/__init__.py: -------------------------------------------------------------------------------- 1 | from .pooling import * 2 | from .embeddings import * 3 | from .blocks import * 4 | from .interactions import * 5 | 6 | 7 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/blocks/__init__.py: -------------------------------------------------------------------------------- 1 | from .logistic_regression import * 2 | from .factorization_machine import * 3 | from .linear import * 4 | from .mlp_block import * 5 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/blocks/factorization_machine.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | import tensorflow as tf 20 | from tensorflow.keras.layers import Layer 21 | from .logistic_regression import LogisticRegression 22 | from ..interactions import InnerProductInteraction 23 | 24 | 25 | class FactorizationMachine(Layer): 26 | def __init__(self, feature_map, regularizer=None): 27 | super(FactorizationMachine, self).__init__() 28 | self.fm_layer = InnerProductInteraction(feature_map.num_fields, output="product_sum") 29 | self.lr_layer = LogisticRegression(feature_map, use_bias=True, regularizer=regularizer) 30 | 31 | def call(self, X, feature_emb): 32 | lr_out = self.lr_layer(X) 33 | fm_out = self.fm_layer(feature_emb) 34 | output = fm_out + lr_out 35 | return output 36 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/blocks/linear.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2023. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | from fuxictr.tensorflow.tf_utils import get_initializer, get_regularizer 20 | from tensorflow.keras.layers import Layer, Dense 21 | 22 | 23 | class Linear(Layer): 24 | def __init__(self, 25 | output_dim, 26 | use_bias=True, 27 | initializer="glorot_normal", 28 | regularizer=None): 29 | super(Linear, self).__init__() 30 | self.linear = Dense(output_dim, use_bias=use_bias, 31 | kernel_initializer=get_initializer(initializer), 32 | kernel_regularizer=get_regularizer(regularizer), 33 | bias_regularizer=get_regularizer(regularizer)) 34 | 35 | def call(self, inputs): 36 | return self.linear(inputs) 37 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/blocks/logistic_regression.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | 19 | import tensorflow as tf 20 | from tensorflow.keras.layers import Layer 21 | from fuxictr.tensorflow.layers import FeatureEmbedding 22 | 23 | 24 | class LogisticRegression(Layer): 25 | def __init__(self, feature_map, use_bias=True, regularizer=None): 26 | super(LogisticRegression, self).__init__() 27 | self.bias = tf.Variable(tf.zeros(1)) if use_bias else None 28 | self.embedding_layer = FeatureEmbedding(feature_map, 1, use_pretrain=False, 29 | use_sharing=False, 30 | embedding_regularizer=regularizer, 31 | name_prefix="lr_") 32 | 33 | def call(self, X): 34 | embed_weights = self.embedding_layer(X) 35 | output = tf.reduce_sum(embed_weights, axis=1) 36 | if self.bias is not None: 37 | output += self.bias 38 | return output 39 | 40 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/embeddings/__init__.py: -------------------------------------------------------------------------------- 1 | from .feature_embedding import * 2 | 3 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/interactions/__init__.py: -------------------------------------------------------------------------------- 1 | from .inner_product import * 2 | from .cross_net import * 3 | 4 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/layers/pooling.py: -------------------------------------------------------------------------------- 1 | # ========================================================================= 2 | # Copyright (C) 2024. The FuxiCTR Library. All rights reserved. 3 | # Copyright (C) 2022. Huawei Technologies Co., Ltd. All rights reserved. 4 | # 5 | # Licensed under the Apache License, Version 2.0 (the "License"); 6 | # you may not use this file except in compliance with the License. 7 | # You may obtain a copy of the License at 8 | # 9 | # http://www.apache.org/licenses/LICENSE-2.0 10 | # 11 | # Unless required by applicable law or agreed to in writing, software 12 | # distributed under the License is distributed on an "AS IS" BASIS, 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | # See the License for the specific language governing permissions and 15 | # limitations under the License. 16 | # ========================================================================= 17 | 18 | import tensorflow as tf 19 | from tensorflow.keras import Model 20 | 21 | 22 | class MaskedSumPooling(Model): 23 | def __init__(self): 24 | super(MaskedSumPooling, self).__init__() 25 | 26 | def forward(self, embedding_matrix): 27 | return tf.reduce_sum(embedding_matrix, axis=1) 28 | 29 | 30 | -------------------------------------------------------------------------------- /fuxictr/tensorflow/models/__init__.py: -------------------------------------------------------------------------------- 1 | from .rank_model import BaseModel -------------------------------------------------------------------------------- /fuxictr/version.py: -------------------------------------------------------------------------------- 1 | __version__="2.3.8" 2 | -------------------------------------------------------------------------------- /model_zoo/AFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/AFM/README.md -------------------------------------------------------------------------------- /model_zoo/AFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/AFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | AFM_test: 16 | model: AFM 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | attention_dim: 8 28 | attention_dropout: [0, 0] 29 | use_attention: True 30 | epochs: 1 31 | shuffle: True 32 | seed: 2019 33 | monitor: 'AUC' 34 | monitor_mode: 'max' 35 | 36 | AFM_default: # This is a config template 37 | model: AFM 38 | dataset_id: TBD 39 | loss: binary_crossentropy 40 | metrics: ['logloss', 'AUC'] 41 | task: binary_classification 42 | optimizer: adam 43 | learning_rate: 1.e-3 44 | embedding_regularizer: 0 45 | net_regularizer: 0 46 | batch_size: 10000 47 | embedding_dim: 40 48 | attention_dim: 40 49 | attention_dropout: [0, 0] 50 | embedding_dropout: 0 51 | use_attention: True 52 | epochs: 100 53 | shuffle: True 54 | seed: 2019 55 | monitor: {'AUC': 1, 'logloss': -1} 56 | monitor_mode: 'max' 57 | 58 | -------------------------------------------------------------------------------- /model_zoo/AFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/AFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .AFM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/AFN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/AFN/README.md -------------------------------------------------------------------------------- /model_zoo/AFN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/AFN/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | AFN_test: 16 | model: AFN 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | logarithmic_neurons: 8 28 | afn_hidden_units: [32, 16] 29 | afn_activations: relu 30 | afn_dropout: 0 31 | ensemble_dnn: True 32 | dnn_hidden_units: [32, 16] 33 | dnn_activations: relu 34 | dnn_dropout: 0 35 | batch_norm: False 36 | epochs: 1 37 | shuffle: True 38 | seed: 2019 39 | monitor: 'AUC' 40 | monitor_mode: 'max' 41 | 42 | AFN_default: # This is a config template 43 | model: AFN 44 | dataset_id: TBD 45 | loss: binary_crossentropy 46 | metrics: ['logloss', 'AUC'] 47 | task: binary_classification 48 | optimizer: adam 49 | learning_rate: 1.e-3 50 | embedding_regularizer: 0 51 | net_regularizer: 0 52 | batch_size: 10000 53 | embedding_dim: 20 54 | logarithmic_neurons: 1200 55 | afn_hidden_units: [400, 400, 400] 56 | afn_activations: relu 57 | afn_dropout: 0 58 | ensemble_dnn: False 59 | dnn_hidden_units: [400, 400, 400] 60 | dnn_activations: relu 61 | dnn_dropout: 0 62 | batch_norm: False 63 | epochs: 100 64 | shuffle: True 65 | seed: 2019 66 | monitor: {'AUC': 1, 'logloss': -1} 67 | monitor_mode: 'max' 68 | 69 | -------------------------------------------------------------------------------- /model_zoo/AFN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/AFN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .AFN import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/AOANet/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/AOANet/README.md -------------------------------------------------------------------------------- /model_zoo/AOANet/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/AOANet/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | AOANet_test: 16 | model: AOANet 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 64, 64] 28 | dnn_hidden_activations: "ReLU" 29 | num_interaction_layers: 3 30 | num_subspaces: 4 31 | batch_norm: False 32 | net_dropout: 0 33 | epochs: 1 34 | shuffle: True 35 | seed: 2022 36 | monitor: 'AUC' 37 | monitor_mode: 'max' 38 | 39 | AOANet_default: # This is a config template 40 | model: AOANet 41 | dataset_id: TBD # TBD is a place holder for assigning dataset_id 42 | loss: 'binary_crossentropy' 43 | metrics: ['logloss', 'AUC'] 44 | task: binary_classification 45 | optimizer: adam 46 | learning_rate: 1.e-3 47 | embedding_regularizer: 0 48 | net_regularizer: 0 49 | batch_size: 4096 50 | embedding_dim: 40 51 | dnn_hidden_units: [64, 64, 64] 52 | dnn_hidden_activations: "ReLU" 53 | num_interaction_layers: 3 54 | num_subspaces: 4 55 | batch_norm: False 56 | net_dropout: 0 57 | epochs: 100 58 | shuffle: True 59 | seed: 2022 60 | monitor: 'AUC' 61 | monitor_mode: 'max' 62 | 63 | -------------------------------------------------------------------------------- /model_zoo/AOANet/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/AOANet/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .AOANet import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/APG/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/APG/README.md -------------------------------------------------------------------------------- /model_zoo/APG/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/APG/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/APG/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .APG_DCNv2 import * 2 | from .APG_DeepFM import * 3 | -------------------------------------------------------------------------------- /model_zoo/AutoInt/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/AutoInt/README.md -------------------------------------------------------------------------------- /model_zoo/AutoInt/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/AutoInt/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | AutoInt_test: 16 | model: AutoInt 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | net_dropout: 0 30 | num_heads: 2 31 | attention_layers: 3 32 | attention_dim: 8 33 | use_residual: True 34 | batch_norm: False 35 | layer_norm: False 36 | use_scale: False 37 | use_wide: False 38 | epochs: 1 39 | shuffle: True 40 | seed: 2019 41 | monitor: 'AUC' 42 | monitor_mode: 'max' 43 | 44 | AutoInt_default: 45 | model: AutoInt 46 | dataset_id: TBD 47 | loss: 'binary_crossentropy' 48 | metrics: ['logloss', 'AUC'] 49 | task: binary_classification 50 | optimizer: adam 51 | learning_rate: 1.e-3 52 | embedding_regularizer: 0 53 | net_regularizer: 0 54 | batch_size: 10000 55 | embedding_dim: 40 56 | dnn_hidden_units: [400, 400] 57 | dnn_activations: relu 58 | net_dropout: 0 59 | num_heads: 2 60 | attention_layers: 3 61 | attention_dim: 40 62 | use_residual: True 63 | batch_norm: False 64 | layer_norm: False 65 | use_scale: False 66 | use_wide: False 67 | epochs: 100 68 | shuffle: True 69 | seed: 2019 70 | monitor: {'AUC': 1, 'logloss': -1} 71 | monitor_mode: 'max' 72 | 73 | -------------------------------------------------------------------------------- /model_zoo/AutoInt/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/AutoInt/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .AutoInt import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/BST/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/BST/README.md -------------------------------------------------------------------------------- /model_zoo/BST/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_seq: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_seq/train.npz 6 | valid_data: ../../data/tiny_seq/valid.npz 7 | test_data: ../../data/tiny_seq/test.npz 8 | 9 | 10 | -------------------------------------------------------------------------------- /model_zoo/BST/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/BST/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .BST import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/CCPM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/CCPM/README.md -------------------------------------------------------------------------------- /model_zoo/CCPM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/CCPM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | CCPM_test: 16 | model: CCPM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | channels: [3, 3] 28 | kernel_heights: [3, 3] 29 | activation: Tanh 30 | epochs: 1 31 | shuffle: True 32 | seed: 2019 33 | monitor: 'AUC' 34 | monitor_mode: 'max' 35 | 36 | CCPM_default: # This is a config template 37 | model: CCPM 38 | dataset_id: TBD 39 | loss: 'binary_crossentropy' 40 | metrics: ['logloss', 'AUC'] 41 | task: binary_classification 42 | optimizer: adam 43 | learning_rate: 1.e-3 44 | embedding_regularizer: 0 45 | net_regularizer: 0 46 | batch_size: 10000 47 | embedding_dim: 40 48 | channels: [3, 3] 49 | kernel_heights: [3, 3] 50 | activation: Tanh 51 | epochs: 100 52 | shuffle: True 53 | seed: 2019 54 | monitor: {'AUC': 1, 'logloss': -1} 55 | monitor_mode: 'max' 56 | 57 | -------------------------------------------------------------------------------- /model_zoo/CCPM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/CCPM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .CCPM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_tf/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DCN/DCN_tf/README.md -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_tf/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_tfrecord: 3 | data_root: ../../../data/ 4 | data_format: tfrecord 5 | train_data: ../../../data/tiny_tfrecord/train.tfrecord 6 | valid_data: ../../../data/tiny_tfrecord/valid.tfrecord 7 | test_data: ../../../data/tiny_tfrecord/test.tfrecord 8 | 9 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_tf/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DCN_test: 16 | model: DCN 17 | dataset_id: tiny_tfrecord 18 | loss: 'binary_crossentropy' 19 | metrics: ['AUC', 'logloss'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 8 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | crossing_layers: 3 30 | net_dropout: 0 31 | batch_norm: False 32 | epochs: 1 33 | shuffle: True 34 | seed: 2019 35 | monitor: 'AUC' 36 | monitor_mode: 'max' 37 | 38 | DCN_default: # This is a config template 39 | model: DCN 40 | dataset_id: TBD 41 | loss: 'binary_crossentropy' 42 | metrics: ['AUC', 'logloss'] 43 | task: binary_classification 44 | optimizer: adam 45 | learning_rate: 1.0e-3 46 | embedding_regularizer: 0 47 | net_regularizer: 0 48 | batch_size: 10000 49 | embedding_dim: 40 50 | dnn_hidden_units: [500, 500, 500] 51 | dnn_activations: relu 52 | num_cross_layers: 3 53 | net_dropout: 0 54 | batch_norm: False 55 | epochs: 100 56 | shuffle: True 57 | seed: 2019 58 | monitor: {'AUC': 1, 'logloss': -1} 59 | monitor_mode: 'max' 60 | 61 | 62 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_tf/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.0.0" 4 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_tf/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DCN import * 2 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_torch/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../../data/ 4 | data_format: npz 5 | train_data: ../../../data/tiny_npz/train.npz 6 | valid_data: ../../../data/tiny_npz/valid.npz 7 | test_data: ../../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_torch/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DCN_test: 16 | model: DCN 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | crossing_layers: 3 30 | net_dropout: 0 31 | batch_norm: False 32 | epochs: 1 33 | shuffle: True 34 | seed: 2019 35 | monitor: 'AUC' 36 | monitor_mode: 'max' 37 | 38 | DCN_default: # This is a config template 39 | model: DCN 40 | dataset_id: TBD 41 | loss: 'binary_crossentropy' 42 | metrics: ['logloss', 'AUC'] 43 | task: binary_classification 44 | optimizer: adam 45 | learning_rate: 1.0e-3 46 | embedding_regularizer: 0 47 | net_regularizer: 0 48 | batch_size: 10000 49 | embedding_dim: 32 50 | dnn_hidden_units: [1024, 512, 256] 51 | dnn_activations: relu 52 | num_cross_layers: 3 53 | net_dropout: 0 54 | batch_norm: False 55 | epochs: 100 56 | shuffle: True 57 | seed: 20222023 58 | monitor: {'AUC': 1, 'logloss': -1} 59 | monitor_mode: 'max' 60 | 61 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_torch/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DCN/DCN_torch/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DCN import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DCNv2/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DCNv2/README.md -------------------------------------------------------------------------------- /model_zoo/DCNv2/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DCNv2/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DCNv2/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DCNv2 import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DESTINE/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DESTINE/README.md -------------------------------------------------------------------------------- /model_zoo/DESTINE/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DESTINE/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DESTINE/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DESTINE import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DIEN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DIEN/README.md -------------------------------------------------------------------------------- /model_zoo/DIEN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_seq: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_seq/train.npz 6 | valid_data: ../../data/tiny_seq/valid.npz 7 | test_data: ../../data/tiny_seq/test.npz 8 | 9 | 10 | -------------------------------------------------------------------------------- /model_zoo/DIEN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DIEN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DIEN import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DIN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DIN/README.md -------------------------------------------------------------------------------- /model_zoo/DIN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_seq: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_seq/train.npz 6 | valid_data: ../../data/tiny_seq/valid.npz 7 | test_data: ../../data/tiny_seq/test.npz 8 | 9 | 10 | -------------------------------------------------------------------------------- /model_zoo/DIN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DIN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DIN import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DLRM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DLRM/README.md -------------------------------------------------------------------------------- /model_zoo/DLRM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DLRM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DLRM_test: 16 | model: DLRM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | top_mlp_units: [64, 64, 64] 28 | bottom_mlp_units: [64, 64, 64] 29 | top_mlp_activations: ReLU 30 | bottom_mlp_activations: ReLU 31 | top_mlp_dropout: 0 32 | bottom_mlp_dropout: 0 33 | interaction_op: dot 34 | batch_norm: False 35 | epochs: 1 36 | shuffle: True 37 | seed: 2022 38 | monitor: 'AUC' 39 | monitor_mode: 'max' 40 | 41 | DLRM_default: # This is a config template 42 | model: DLRM 43 | dataset_id: TBD 44 | loss: 'binary_crossentropy' 45 | metrics: ['logloss', 'AUC'] 46 | task: binary_classification 47 | optimizer: adam 48 | learning_rate: 1.e-3 49 | embedding_regularizer: 0 50 | net_regularizer: 0 51 | batch_size: 4096 52 | embedding_dim: 40 53 | top_mlp_units: [64, 64, 64] 54 | bottom_mlp_units: [64, 64, 64] 55 | top_mlp_activations: ReLU 56 | bottom_mlp_activations: ReLU 57 | top_mlp_dropout: 0 58 | bottom_mlp_dropout: 0 59 | interaction_op: dot 60 | batch_norm: False 61 | epochs: 100 62 | shuffle: True 63 | seed: 2022 64 | monitor: 'AUC' 65 | monitor_mode: 'max' 66 | 67 | -------------------------------------------------------------------------------- /model_zoo/DLRM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DLRM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DLRM import * 2 | -------------------------------------------------------------------------------- /model_zoo/DMIN/README.md: -------------------------------------------------------------------------------- 1 | ## DMIN 2 | 3 | Zhibo Xiao, Luwei Yang, Wen Jiang, Yi Wei, Yi Hu, Hao Wang. Deep Multi-Interest Network for Click-through Rate Prediction, in CIKM 2020. 4 | 5 | 6 | -------------------------------------------------------------------------------- /model_zoo/DMIN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_seq: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_seq/train.npz 6 | valid_data: ../../data/tiny_seq/valid.npz 7 | test_data: ../../data/tiny_seq/test.npz 8 | 9 | 10 | -------------------------------------------------------------------------------- /model_zoo/DMIN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DMIN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DMIN import * 2 | -------------------------------------------------------------------------------- /model_zoo/DMR/README.md: -------------------------------------------------------------------------------- 1 | ## DMR 2 | 3 | Ze Lyu, Yu Dong, Chengfu Huo, Weijun Ren. Deep Match to Rank Model for Personalized Click-Through Rate Prediction, in AAAI 2020. 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DMR/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_seq: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_seq/train.npz 6 | valid_data: ../../data/tiny_seq/valid.npz 7 | test_data: ../../data/tiny_seq/test.npz 8 | 9 | 10 | -------------------------------------------------------------------------------- /model_zoo/DMR/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DMR/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DMR import * 2 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_tf/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DNN/DNN_tf/README.md -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_tf/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_tfrecord: 3 | data_root: ../../../data/ 4 | data_format: tfrecord 5 | train_data: ../../../data/tiny_tfrecord/train.tfrecord 6 | valid_data: ../../../data/tiny_tfrecord/valid.tfrecord 7 | test_data: ../../../data/tiny_tfrecord/test.tfrecord 8 | 9 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_tf/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DNN_test: 16 | model: DNN 17 | dataset_id: tiny_tfrecord 18 | loss: 'binary_crossentropy' 19 | metrics: ['AUC', 'logloss'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 8 26 | embedding_dim: 4 27 | hidden_units: [64, 32] 28 | hidden_activations: relu 29 | net_dropout: 0 30 | batch_norm: False 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | DNN_default: # This is a config template 38 | model: DCN 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['AUC', 'logloss'] 42 | task: binary_classification 43 | optimizer: adam 44 | learning_rate: 1.e-3 45 | embedding_regularizer: 0 46 | net_regularizer: 0 47 | batch_size: 10000 48 | embedding_dim: 40 49 | hidden_units: [500, 500, 500] 50 | hidden_activations: relu 51 | net_dropout: 0 52 | batch_norm: False 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | 59 | 60 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_tf/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.0.0" 4 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_tf/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DNN import DNN 2 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_torch/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DNN/DNN_torch/README.md -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_torch/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../../data/ 4 | data_format: npz 5 | train_data: ../../../data/tiny_npz/train.npz 6 | valid_data: ../../../data/tiny_npz/valid.npz 7 | test_data: ../../../data/tiny_npz/test.npz 8 | 9 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_torch/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DNN_test: 16 | model: DNN 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | hidden_units: [64, 32] 28 | hidden_activations: relu 29 | net_dropout: 0 30 | batch_norm: False 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | DNN_default: # This is a config template 38 | model: DNN 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | learning_rate: 1.e-3 45 | embedding_regularizer: 0 46 | net_regularizer: 0 47 | batch_size: 10000 48 | embedding_dim: 40 49 | hidden_units: [500, 500, 500] 50 | hidden_activations: relu 51 | net_dropout: 0 52 | batch_norm: False 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | 59 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_torch/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DNN/DNN_torch/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DNN import DNN 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DSSM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DSSM/README.md -------------------------------------------------------------------------------- /model_zoo/DSSM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DSSM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DSSM_test: 16 | model: DSSM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | user_tower_units: [64, 64, 64] 28 | item_tower_units: [64, 64, 64] 29 | user_tower_activations: "ReLU" 30 | item_tower_activations: "ReLU" 31 | user_tower_dropout: 0 32 | item_tower_dropout: 0 33 | batch_norm: False 34 | epochs: 1 35 | shuffle: True 36 | seed: 2022 37 | monitor: 'AUC' 38 | monitor_mode: 'max' 39 | 40 | DSSM_default: # This is a config template 41 | model: DSSM 42 | dataset_id: TBD 43 | loss: 'binary_crossentropy' 44 | metrics: ['logloss', 'AUC'] 45 | task: binary_classification 46 | optimizer: adam 47 | learning_rate: 1.e-3 48 | embedding_regularizer: 0 49 | net_regularizer: 0 50 | batch_size: 4096 51 | embedding_dim: 40 52 | user_tower_units: [64, 64, 64] 53 | item_tower_units: [64, 64, 64] 54 | user_tower_activations: "ReLU" 55 | item_tower_activations: "ReLU" 56 | user_tower_dropout: 0 57 | item_tower_dropout: 0 58 | batch_norm: False 59 | epochs: 100 60 | shuffle: True 61 | seed: 2022 62 | monitor: 'AUC' 63 | monitor_mode: 'max' 64 | 65 | -------------------------------------------------------------------------------- /model_zoo/DSSM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DSSM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DSSM import * 2 | -------------------------------------------------------------------------------- /model_zoo/DeepCrossing/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DeepCrossing/README.md -------------------------------------------------------------------------------- /model_zoo/DeepCrossing/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DeepCrossing/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepCrossing_test: 16 | model: DeepCrossing 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | residual_blocks: [64, 64, 64] 28 | dnn_activations: relu 29 | net_dropout: 0 30 | batch_norm: False 31 | use_residual: True 32 | epochs: 1 33 | shuffle: True 34 | seed: 2019 35 | monitor: 'AUC' 36 | monitor_mode: 'max' 37 | 38 | DeepCrossing_default: # This is a config template 39 | model: DeepCrossing 40 | dataset_id: TBD 41 | loss: 'binary_crossentropy' 42 | metrics: ['logloss', 'AUC'] 43 | task: binary_classification 44 | optimizer: adam 45 | learning_rate: 1.0e-3 46 | embedding_regularizer: 0 47 | net_regularizer: 0 48 | batch_size: 10000 49 | embedding_dim: 40 50 | residual_blocks: [500, 500, 500] 51 | dnn_activations: relu 52 | net_dropout: 0 53 | batch_norm: False 54 | use_residual: True 55 | epochs: 100 56 | shuffle: True 57 | seed: 2019 58 | monitor: {'AUC': 1, 'logloss': -1} 59 | monitor_mode: 'max' 60 | 61 | -------------------------------------------------------------------------------- /model_zoo/DeepCrossing/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DeepCrossing/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DeepCrossing import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_tf/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DeepFM/DeepFM_tf/README.md -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_tf/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_tfrecord: 3 | data_root: ../../../data/ 4 | data_format: tfrecord 5 | train_data: ../../../data/tiny_tfrecord/train.tfrecord 6 | valid_data: ../../../data/tiny_tfrecord/valid.tfrecord 7 | test_data: ../../../data/tiny_tfrecord/test.tfrecord 8 | 9 | -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_tf/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepFM_test: 16 | model: DeepFM 17 | dataset_id: tiny_tfrecord 18 | loss: 'binary_crossentropy' 19 | metrics: ['AUC', 'logloss'] 20 | task: binary_classification 21 | optimizer: adam 22 | hidden_units: [64, 32] 23 | hidden_activations: relu 24 | net_regularizer: 0 25 | embedding_regularizer: 1.e-8 26 | learning_rate: 1.e-3 27 | batch_norm: False 28 | net_dropout: 0 29 | batch_size: 8 30 | embedding_dim: 4 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: {'AUC': 1, 'logloss': -1} 35 | monitor_mode: 'max' 36 | 37 | DeepFM_default: # This is a config template 38 | model: DeepFM 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['AUC', 'logloss'] 42 | task: binary_classification 43 | optimizer: adam 44 | hidden_units: [300, 300, 300] 45 | hidden_activations: relu 46 | net_regularizer: 0 47 | embedding_regularizer: 0 48 | learning_rate: 1.e-3 49 | batch_norm: False 50 | net_dropout: 0 51 | batch_size: 10000 52 | embedding_dim: 40 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | 59 | -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_tf/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.0.0" 4 | -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_tf/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DeepFM import * 2 | -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_torch/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DeepFM/DeepFM_torch/README.md -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_torch/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_torch/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DeepFM/DeepFM_torch/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DeepFM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/DeepIM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/DeepIM/README.md -------------------------------------------------------------------------------- /model_zoo/DeepIM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/DeepIM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DeepIM_test: 16 | model: DeepIM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | im_order: 2 23 | im_batch_norm: False 24 | hidden_units: [64, 32] 25 | hidden_activations: relu 26 | net_regularizer: 0 27 | embedding_regularizer: 0 28 | learning_rate: 1.e-3 29 | net_batch_norm: False 30 | net_dropout: 0 31 | batch_size: 128 32 | embedding_dim: 4 33 | epochs: 1 34 | shuffle: True 35 | seed: 2019 36 | monitor: 'AUC' 37 | monitor_mode: 'max' 38 | 39 | DeepIM_default: # This is a config template 40 | model: DeepIM 41 | dataset_id: TBD 42 | loss: 'binary_crossentropy' 43 | metrics: ['logloss', 'AUC'] 44 | task: binary_classification 45 | optimizer: adam 46 | im_order: 2 47 | im_batch_norm: False 48 | hidden_units: [300, 300, 300] 49 | hidden_activations: relu 50 | net_regularizer: 0 51 | embedding_regularizer: 0 52 | learning_rate: 1.e-3 53 | net_batch_norm: False 54 | net_dropout: 0 55 | batch_size: 10000 56 | embedding_dim: 40 57 | epochs: 100 58 | shuffle: True 59 | seed: 2019 60 | monitor: {'AUC': 1, 'logloss': -1} 61 | monitor_mode: 'max' 62 | 63 | -------------------------------------------------------------------------------- /model_zoo/DeepIM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/DeepIM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DeepIM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/EDCN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/EDCN/README.md -------------------------------------------------------------------------------- /model_zoo/EDCN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/EDCN/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | EDCN_test: 16 | model: EDCN 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | num_cross_layers: 2 28 | hidden_activations: ReLU 29 | bridge_type: "hadamard_product" 30 | temperature: 1 31 | batch_norm: False 32 | net_dropout: 0 33 | epochs: 1 34 | shuffle: True 35 | seed: 2022 36 | monitor: 'AUC' 37 | monitor_mode: 'max' 38 | 39 | EDCN_default: # This is a config template 40 | model: EDCN 41 | dataset_id: TBD 42 | loss: 'binary_crossentropy' 43 | metrics: ['logloss', 'AUC'] 44 | task: binary_classification 45 | optimizer: adam 46 | learning_rate: 1.e-3 47 | embedding_regularizer: 0 48 | net_regularizer: 0 49 | batch_size: 4096 50 | embedding_dim: 40 51 | num_cross_layers: 3 52 | hidden_activations: ReLU 53 | bridge_type: "hadamard_product" 54 | temperature: 1 55 | batch_norm: False 56 | net_dropout: 0 57 | epochs: 100 58 | shuffle: True 59 | seed: 2022 60 | monitor: 'AUC' 61 | monitor_mode: 'max' 62 | 63 | -------------------------------------------------------------------------------- /model_zoo/EDCN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/EDCN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .EDCN import * 2 | -------------------------------------------------------------------------------- /model_zoo/EulerNet/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz -------------------------------------------------------------------------------- /model_zoo/EulerNet/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | EulerNet_default: # This is a config template 16 | model: EulerNet 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | shape: [20] 23 | net_ex_dropout: 0.0 24 | net_im_dropout: 0.0 25 | layer_norm: true 26 | learning_rate: 1.0e-3 27 | embedding_regularizer: 0 28 | net_regularizer: 0 29 | batch_size: 10000 30 | embedding_dim: 10 31 | epochs: 100 32 | shuffle: True 33 | seed: 2021 34 | monitor: {'AUC': 1, 'logloss': -1} 35 | monitor_mode: 'max' 36 | 37 | EulerNet_test: 38 | model: EulerNet 39 | dataset_id: tiny_npz 40 | loss: 'binary_crossentropy' 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | shape: [52] 45 | net_ex_dropout: 0.1 46 | net_im_dropout: 0.1 47 | layer_norm: true 48 | learning_rate: 1.0e-3 49 | embedding_regularizer: 0 50 | net_regularizer: 0 51 | batch_size: 10000 52 | embedding_dim: 10 53 | epochs: 1 54 | shuffle: True 55 | seed: 2021 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | -------------------------------------------------------------------------------- /model_zoo/EulerNet/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/EulerNet/readme.md: -------------------------------------------------------------------------------- 1 | ## Tuning 2 | The detailed tuning config files are provided in this repo: https://github.com/Ethan-TZ/EulerNet/tree/main/%23Code4FuxiCTR%23. 3 | 4 | ## Cite 5 | 6 | If you find EulerNet useful for your research or development, please cite the following papers: [EulerNet](https://dl.acm.org/doi/10.1145/3539618.3591681). 7 | 8 | ```bibtex 9 | @inproceedings{tian2023eulernet, 10 | title = {EulerNet: Adaptive Feature Interaction Learning via Euler's Formula for CTR Prediction}, 11 | author = {Tian, Zhen and Bai, Ting and Zhao, Wayne Xin and Wen, Ji-Rong and Cao, Zhao}, 12 | booktitle = {Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval}, 13 | pages = {1376–1385}, 14 | year = {2023}, 15 | } 16 | ``` 17 | -------------------------------------------------------------------------------- /model_zoo/EulerNet/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .EulerNet import * -------------------------------------------------------------------------------- /model_zoo/FFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FFM/README.md -------------------------------------------------------------------------------- /model_zoo/FFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FFM import * 2 | from .FFMv2 import * 3 | 4 | 5 | 6 | -------------------------------------------------------------------------------- /model_zoo/FGCNN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FGCNN/README.md -------------------------------------------------------------------------------- /model_zoo/FGCNN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FGCNN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FGCNN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FGCNN import * 2 | -------------------------------------------------------------------------------- /model_zoo/FLEN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FLEN/README.md -------------------------------------------------------------------------------- /model_zoo/FLEN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FLEN/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | FLEN_test: 16 | model: FLEN 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | net_dropout: 0 30 | batch_norm: False 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | FLEN_default: # This is a config template 38 | model: FLEN 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | learning_rate: 1.0e-3 45 | embedding_regularizer: 0 46 | net_regularizer: 0 47 | batch_size: 10000 48 | embedding_dim: 40 49 | dnn_hidden_units: [500, 500, 500] 50 | dnn_activations: relu 51 | net_dropout: 0 52 | batch_norm: False 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | 59 | 60 | -------------------------------------------------------------------------------- /model_zoo/FLEN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FLEN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FLEN import * 2 | -------------------------------------------------------------------------------- /model_zoo/FM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FM/README.md -------------------------------------------------------------------------------- /model_zoo/FM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | FM_test: 16 | model: FM 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | regularizer: 1.e-8 24 | batch_size: 128 25 | embedding_dim: 4 26 | epochs: 1 27 | shuffle: True 28 | seed: 2019 29 | monitor: 'AUC' 30 | monitor_mode: 'max' 31 | 32 | FM_default: # This is a config template 33 | model: FM 34 | dataset_id: TBD 35 | loss: binary_crossentropy 36 | metrics: ['logloss', 'AUC'] 37 | task: binary_classification 38 | optimizer: adam 39 | learning_rate: 1.0e-3 40 | regularizer: 0 41 | batch_size: 10000 42 | embedding_dim: 10 43 | epochs: 100 44 | shuffle: True 45 | seed: 2019 46 | monitor: {'AUC': 1, 'logloss': -1} 47 | monitor_mode: 'max' 48 | -------------------------------------------------------------------------------- /model_zoo/FM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FM import FM -------------------------------------------------------------------------------- /model_zoo/FiBiNET/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FiBiNET/README.md -------------------------------------------------------------------------------- /model_zoo/FiBiNET/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FiBiNET/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | FiBiNET_test: 16 | model: FiBiNET 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | hidden_units: [64, 32] 28 | hidden_activations: relu 29 | bilinear_type: field_interaction 30 | reduction_ratio: 3 31 | net_dropout: 0 32 | batch_norm: False 33 | epochs: 1 34 | shuffle: True 35 | seed: 2019 36 | monitor: 'AUC' 37 | monitor_mode: 'max' 38 | 39 | FiBiNET_default: # This is a config template 40 | model: FiBiNET 41 | dataset_id: TBD 42 | loss: binary_crossentropy 43 | metrics: ['logloss', 'AUC'] 44 | task: binary_classification 45 | optimizer: adam 46 | learning_rate: 1.0e-3 47 | embedding_regularizer: 0 48 | net_regularizer: 0 49 | batch_size: 10000 50 | embedding_dim: 40 51 | hidden_units: [500, 500, 500] 52 | hidden_activations: relu 53 | bilinear_type: field_interaction 54 | reduction_ratio: 3 55 | net_dropout: 0 56 | batch_norm: False 57 | epochs: 100 58 | shuffle: True 59 | seed: 2019 60 | monitor: {'AUC': 1, 'logloss': -1} 61 | monitor_mode: 'max' 62 | -------------------------------------------------------------------------------- /model_zoo/FiBiNET/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FiBiNET/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FiBiNET import * 2 | -------------------------------------------------------------------------------- /model_zoo/FiGNN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FiGNN/README.md -------------------------------------------------------------------------------- /model_zoo/FiGNN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FiGNN/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | FiGNN_test: 16 | model: FiGNN 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | gnn_layers: 2 28 | use_residual: True 29 | use_gru: True 30 | reuse_graph_layer: False 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | FiGNN_default: # This is a config template 38 | model: FiGNN 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | learning_rate: 1.e-3 45 | embedding_regularizer: 0 46 | net_regularizer: 0 47 | batch_size: 10000 48 | embedding_dim: 40 49 | gnn_layers: 3 50 | use_residual: True 51 | use_gru: True 52 | reuse_graph_layer: False 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | -------------------------------------------------------------------------------- /model_zoo/FiGNN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FiGNN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FiGNN import * 2 | -------------------------------------------------------------------------------- /model_zoo/FinalMLP/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FinalMLP/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FinalMLP/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .DualMLP import * 2 | from .FinalMLP import * -------------------------------------------------------------------------------- /model_zoo/FinalNet/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FinalNet/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FinalNet/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FinalNet import * -------------------------------------------------------------------------------- /model_zoo/FmFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FmFM/README.md -------------------------------------------------------------------------------- /model_zoo/FmFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FmFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | FmFM_test: 16 | model: FmFM 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | regularizer: 0 24 | field_interaction_type: matrixed 25 | batch_size: 128 26 | embedding_dim: 4 27 | epochs: 1 28 | shuffle: True 29 | seed: 2019 30 | monitor: 'AUC' 31 | monitor_mode: 'max' 32 | 33 | FmFM_default: # This is a config template 34 | model: FmFM 35 | dataset_id: TBD 36 | loss: binary_crossentropy 37 | metrics: ['logloss', 'AUC'] 38 | task: binary_classification 39 | optimizer: adam 40 | learning_rate: 1.0e-3 41 | regularizer: 0 42 | field_interaction_type: matrixed 43 | batch_size: 10000 44 | embedding_dim: 10 45 | epochs: 100 46 | shuffle: True 47 | seed: 2019 48 | monitor: {'AUC': 1, 'logloss': -1} 49 | monitor_mode: 'max' 50 | -------------------------------------------------------------------------------- /model_zoo/FmFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FmFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FmFM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/FwFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/FwFM/README.md -------------------------------------------------------------------------------- /model_zoo/FwFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/FwFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | FwFM_test: 16 | model: FwFM 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | regularizer: 1.e-8 24 | batch_size: 128 25 | embedding_dim: 4 26 | epochs: 1 27 | shuffle: True 28 | seed: 2019 29 | monitor: 'AUC' 30 | monitor_mode: 'max' 31 | 32 | FwFM_default: # This is a config template 33 | model: FwFM 34 | dataset_id: TBD 35 | loss: binary_crossentropy 36 | metrics: ['logloss', 'AUC'] 37 | task: binary_classification 38 | optimizer: adam 39 | learning_rate: 1.0e-3 40 | regularizer: 0 41 | linear_type: FiLV 42 | batch_size: 10000 43 | embedding_dim: 40 44 | epochs: 100 45 | shuffle: True 46 | seed: 2019 47 | monitor: {'AUC': 1, 'logloss': -1} 48 | monitor_mode: 'max' 49 | -------------------------------------------------------------------------------- /model_zoo/FwFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/FwFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .FwFM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/GDCN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | 9 | -------------------------------------------------------------------------------- /model_zoo/GDCN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/GDCN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .GDCN import GDCNP, GDCN 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/HFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/HFM/README.md -------------------------------------------------------------------------------- /model_zoo/HFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/HFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | HFM_test: 16 | model: HFM 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | interaction_type: circular_convolution 24 | use_dnn: False 25 | hidden_units: [64, 32] 26 | hidden_activations: relu 27 | embedding_regularizer: 1.e-8 28 | net_regularizer: 0 29 | net_dropout: 0 30 | batch_norm: False 31 | batch_size: 128 32 | embedding_dim: 4 33 | epochs: 1 34 | shuffle: True 35 | seed: 2019 36 | monitor: 'AUC' 37 | monitor_mode: 'max' 38 | 39 | HFM_default: # This is a config template 40 | model: HFM 41 | dataset_id: TBD 42 | loss: binary_crossentropy 43 | metrics: ['logloss', 'AUC'] 44 | task: binary_classification 45 | optimizer: adam 46 | learning_rate: 1.0e-3 47 | interaction_type: circular_convolution 48 | use_dnn: False 49 | hidden_units: [64, 64, 64] 50 | hidden_activations: relu 51 | embedding_regularizer: 0 52 | net_regularizer: 0 53 | net_dropout: 0 54 | batch_norm: False 55 | batch_size: 10000 56 | embedding_dim: 40 57 | epochs: 100 58 | shuffle: True 59 | seed: 2019 60 | monitor: {'AUC': 1, 'logloss': -1} 61 | monitor_mode: 'max' 62 | -------------------------------------------------------------------------------- /model_zoo/HFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/HFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .HFM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/HOFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/HOFM/README.md -------------------------------------------------------------------------------- /model_zoo/HOFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/HOFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | HOFM_test: 16 | model: HOFM 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | regularizer: 1.e-8 24 | order: 3 25 | batch_size: 128 26 | embedding_dim: 4 27 | reuse_embedding: False 28 | epochs: 1 29 | shuffle: True 30 | seed: 2019 31 | monitor: 'AUC' 32 | monitor_mode: 'max' 33 | 34 | HOFM_default: # This is a config template 35 | model: HOFM 36 | dataset_id: TBD 37 | loss: binary_crossentropy 38 | metrics: ['logloss', 'AUC'] 39 | task: binary_classification 40 | optimizer: adam 41 | learning_rate: 1.0e-3 42 | regularizer: 0 43 | order: 4 44 | batch_size: 10000 45 | embedding_dim: 10 46 | reuse_embedding: False 47 | epochs: 100 48 | shuffle: True 49 | seed: 2019 50 | monitor: {'AUC': 1, 'logloss': -1} 51 | monitor_mode: 'max' 52 | -------------------------------------------------------------------------------- /model_zoo/HOFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/HOFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .HOFM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/InterHAt/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/InterHAt/README.md -------------------------------------------------------------------------------- /model_zoo/InterHAt/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/InterHAt/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | InterHAt_test: 16 | model: InterHAt 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | hidden_dim: null 28 | order: 3 29 | num_heads: 1 30 | attention_dim: 8 31 | hidden_units: [64, 32] 32 | hidden_activations: relu 33 | net_dropout: 0 34 | batch_norm: False 35 | layer_norm: True 36 | use_residual: True 37 | epochs: 1 38 | shuffle: True 39 | seed: 2019 40 | monitor: 'AUC' 41 | monitor_mode: 'max' 42 | 43 | InterHAt_default: # This is a config template 44 | model: InterHAt 45 | dataset_id: TBD 46 | loss: 'binary_crossentropy' 47 | metrics: ['logloss', 'AUC'] 48 | task: binary_classification 49 | optimizer: adam 50 | learning_rate: 1.0e-3 51 | embedding_regularizer: 0 52 | net_regularizer: 0 53 | batch_size: 10000 54 | embedding_dim: 40 55 | hidden_dim: null 56 | order: 3 57 | num_heads: 1 58 | attention_dim: 40 59 | hidden_units: [64] 60 | hidden_activations: relu 61 | net_dropout: 0 62 | batch_norm: False 63 | layer_norm: True 64 | use_residual: True 65 | epochs: 100 66 | shuffle: True 67 | seed: 2019 68 | monitor: {'AUC': 1, 'logloss': -1} 69 | monitor_mode: 'max' 70 | -------------------------------------------------------------------------------- /model_zoo/InterHAt/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/InterHAt/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .InterHAt import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/LR/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/LR/README.md -------------------------------------------------------------------------------- /model_zoo/LR/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/LR/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | LR_test: 16 | model: LR 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | regularizer: 1.e-8 24 | batch_size: 128 25 | epochs: 1 26 | shuffle: True 27 | seed: 2019 28 | monitor: 'AUC' 29 | monitor_mode: 'max' 30 | 31 | LR_default: # This is a config template 32 | model: LR 33 | dataset_id: TBD 34 | loss: binary_crossentropy 35 | metrics: ['logloss', 'AUC'] 36 | task: binary_classification 37 | optimizer: adam 38 | learning_rate: 1.0e-3 39 | regularizer: 0 40 | batch_size: 10000 41 | epochs: 100 42 | shuffle: True 43 | seed: 2019 44 | monitor: {'AUC': 1, 'logloss': -1} 45 | monitor_mode: 'max' 46 | -------------------------------------------------------------------------------- /model_zoo/LR/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/LR/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .LR import LR 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/DCNv2/__init__.py: -------------------------------------------------------------------------------- 1 | from .DCNv2 import DCNv2 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/DIEN/__init__.py: -------------------------------------------------------------------------------- 1 | from .DIEN import DIEN 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/DIN/__init__.py: -------------------------------------------------------------------------------- 1 | from .DIN import DIN 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/ETA/__init__.py: -------------------------------------------------------------------------------- 1 | from .ETA import ETA 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/FinalMLP/__init__.py: -------------------------------------------------------------------------------- 1 | from .FinalMLP import FinalMLP 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/MIRRN/__init__.py: -------------------------------------------------------------------------------- 1 | from .MIRRN import MIRRN 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/SDIM/__init__.py: -------------------------------------------------------------------------------- 1 | from .SDIM import SDIM 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/SIM/__init__.py: -------------------------------------------------------------------------------- 1 | from .SIM import SIM 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/TWIN/__init__.py: -------------------------------------------------------------------------------- 1 | from .TWIN import TWIN 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/TransAct/__init__.py: -------------------------------------------------------------------------------- 1 | from .TransAct import TransAct 2 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/__init__.py: -------------------------------------------------------------------------------- 1 | from .DIN import DIN 2 | from .DCNv2 import DCNv2 3 | from .DIEN import DIEN 4 | from .FinalMLP import FinalMLP 5 | from .ETA import ETA 6 | from .MIRRN import MIRRN 7 | from .TransAct import TransAct 8 | from .SIM import SIM 9 | from .SDIM import SDIM 10 | from .TWIN import TWIN 11 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/dcnv2_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 4 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DCNv2_default: # This is a config template 16 | model: DCNv2 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | model_structure: parallel 23 | use_low_rank_mixture: False 24 | low_rank: 32 25 | num_experts: 4 26 | learning_rate: 1.0e-3 27 | embedding_regularizer: 0 28 | net_regularizer: 0 29 | batch_size: 10000 30 | embedding_dim: 40 31 | stacked_dnn_hidden_units: [500, 500, 500] 32 | parallel_dnn_hidden_units: [500, 500, 500] 33 | dnn_activations: relu 34 | num_cross_layers: 3 35 | net_dropout: 0 36 | batch_norm: False 37 | epochs: 100 38 | shuffle: True 39 | seed: 2019 40 | monitor: {'AUC': 1, 'logloss': -1} 41 | monitor_mode: 'max' 42 | max_len: 50 43 | accumulation_steps: 1 44 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/dien_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DIEN_default: # This is a config template 16 | model: DIEN 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 10000 26 | embedding_dim: 16 27 | dnn_hidden_units: [1024, 512, 256] 28 | dnn_activations: Dice 29 | net_dropout: 0 30 | gru_type: AUGRU 31 | enable_sum_pooling: False 32 | attention_type: bilinear_attention 33 | attention_hidden_units: [80, 40] 34 | attention_activation: "ReLU" 35 | attention_dropout: 0 36 | use_attention_softmax: True 37 | aux_hidden_units: [100, 50] 38 | aux_activation: "ReLU" 39 | aux_loss_alpha: 0 40 | batch_norm: True 41 | epochs: 100 42 | shuffle: True 43 | seed: 20222023 44 | monitor: {'AUC': 1, 'logloss': -1} 45 | monitor_mode: 'max' 46 | max_len: 50 47 | accumulation_steps: 1 48 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/din_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 4 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | DIN_default: # This is a config template 16 | model: DIN 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: [AUC, logloss] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 8192 26 | embedding_dim: 64 27 | dnn_hidden_units: [500, 500, 500] 28 | dnn_activations: relu 29 | attention_hidden_units: [64] 30 | attention_hidden_activations: "Dice" 31 | attention_output_activation: null 32 | attention_dropout: 0 33 | din_use_softmax: False 34 | net_dropout: 0 35 | batch_norm: False 36 | epochs: 100 37 | shuffle: True 38 | seed: 2019 39 | monitor: {'AUC': 1, 'logloss': -1} 40 | monitor_mode: 'max' 41 | max_len: 50 42 | accumulation_steps: 1 43 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/eta_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | ETA_default: # This is a config template 16 | model: ETA 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 8192 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | attention_dim: 64 30 | num_heads: 2 31 | use_scale: True 32 | attention_dropout: 0 33 | reuse_hash: True 34 | hash_bits: 32 35 | topk: 50 36 | short_seq_len: 50 37 | net_dropout: 0 38 | batch_norm: False 39 | epochs: 100 40 | shuffle: True 41 | seed: 20222023 42 | monitor: 'AUC' 43 | monitor_mode: 'max' 44 | max_len: 50 45 | accumulation_steps: 1 46 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/finalmlp_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | FinalMLP_default: # This is a config template 16 | model: FinalMLP 17 | dataset_id: TBD 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | mlp1_hidden_units: [1024, 512] 26 | mlp1_hidden_activations: relu 27 | mlp1_dropout: 0 28 | mlp1_batch_norm: False 29 | mlp2_hidden_units: [1024, 512, 256] 30 | mlp2_hidden_activations: relu 31 | mlp2_dropout: 0 32 | mlp2_batch_norm: False 33 | use_fs: True 34 | fs_hidden_units: [1024, 512] 35 | fs1_context: [] 36 | fs2_context: [] 37 | num_heads: 2 38 | batch_size: 10000 39 | embedding_dim: 16 40 | epochs: 100 41 | shuffle: True 42 | seed: 20222023 43 | monitor: {'AUC': 1, 'logloss': -1} 44 | monitor_mode: 'max' 45 | max_len: 50 46 | accumulation_steps: 1 47 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/mirrn_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 4 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | MIRRN_default: # This is a config template 16 | model: MIRRN 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 256 26 | embedding_dim: 16 27 | # dnn 28 | batch_norm: False 29 | dnn_hidden_units: [200, 80] 30 | dnn_activations: relu 31 | net_dropout: 0 32 | # target attention 33 | attention_dim: 32 34 | num_heads: 4 35 | use_scale: True 36 | attention_dropout: 0.1 37 | short_seq_len: 50 38 | # eta 39 | reuse_hash: True 40 | hash_bits: 32 41 | topk: 4 42 | # training 43 | epochs: 1 44 | shuffle: True 45 | seed: 2024 46 | monitor: 'AUC' 47 | monitor_mode: 'max' 48 | max_len: 50 49 | accumulation_steps: 1 50 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/sdim_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | SDIM_default: # This is a config template 16 | model: SDIM 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 10000 26 | embedding_dim: 32 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | attention_dim: 64 30 | use_qkvo: True 31 | num_heads: 2 32 | use_scale: True 33 | attention_dropout: 0 34 | reuse_hash: True 35 | num_hashes: 2 36 | hash_bits: 4 37 | net_dropout: 0 38 | batch_norm: False 39 | l2_norm: False 40 | epochs: 100 41 | shuffle: True 42 | seed: 20222023 43 | monitor: 'AUC' 44 | monitor_mode: 'max' 45 | short_seq_len: 50 46 | max_len: 50 47 | accumulation_steps: 1 48 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/sim_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | SIM_default: # This is a config template 16 | model: SIM 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 8192 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | attention_dim: 64 30 | num_heads: 2 31 | gsu_type: "soft" 32 | alpha: 1 33 | beta: 1 34 | attention_dropout: 0 35 | topk: 50 36 | short_seq_len: 50 37 | net_dropout: 0 38 | batch_norm: False 39 | epochs: 100 40 | shuffle: True 41 | seed: 20222023 42 | monitor: 'AUC' 43 | monitor_mode: 'max' 44 | max_len: 50 45 | accumulation_steps: 1 46 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/transact_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | TransAct_default: # This is a config template 16 | model: TransAct 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 10000 26 | embedding_dim: 64 27 | hidden_activations: relu 28 | dcn_cross_layers: 3 29 | dcn_hidden_units: [1024, 512, 256] 30 | mlp_hidden_units: [] 31 | num_heads: 1 32 | transformer_layers: 1 33 | transformer_dropout: 0 34 | dim_feedforward: 512 35 | net_dropout: 0 36 | first_k_cols: 1 37 | use_time_window_mask: False 38 | time_window_ms: 86400000 39 | concat_max_pool: True 40 | batch_norm: False 41 | epochs: 100 42 | shuffle: True 43 | seed: 20242025 44 | monitor: {'AUC': 1, 'logloss': -1} 45 | monitor_mode: 'max' 46 | max_len: 50 47 | accumulation_steps: 1 48 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/config/twin_config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | TWIN_default: # This is a config template 16 | model: TWIN 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 8192 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_activations: relu 29 | attention_dim: 64 30 | num_heads: 2 31 | attention_dropout: 0 32 | topk: 50 33 | Kc_cross_features: 0 34 | short_seq_len: 50 35 | net_dropout: 0 36 | batch_norm: False 37 | epochs: 100 38 | shuffle: True 39 | seed: 20222023 40 | monitor: 'AUC' 41 | monitor_mode: 'max' 42 | max_len: 50 43 | accumulation_steps: 1 44 | -------------------------------------------------------------------------------- /model_zoo/LongCTR/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | import fuxictr 2 | assert fuxictr.__version__ == "2.3.2" 3 | -------------------------------------------------------------------------------- /model_zoo/LorentzFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/LorentzFM/README.md -------------------------------------------------------------------------------- /model_zoo/LorentzFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/LorentzFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | LorentzFM_test: 16 | model: LorentzFM 17 | dataset_id: tiny_npz 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | regularizer: 1.e-8 24 | batch_size: 128 25 | embedding_dim: 4 26 | epochs: 1 27 | shuffle: True 28 | seed: 2019 29 | monitor: 'AUC' 30 | monitor_mode: 'max' 31 | 32 | LorentzFM_default: # This is a config template 33 | model: LorentzFM 34 | dataset_id: TBD 35 | loss: binary_crossentropy 36 | metrics: ['logloss', 'AUC'] 37 | task: binary_classification 38 | optimizer: adam 39 | learning_rate: 1.0e-3 40 | regularizer: 0 41 | batch_size: 10000 42 | embedding_dim: 40 43 | epochs: 100 44 | shuffle: True 45 | seed: 2019 46 | monitor: {'AUC': 1, 'logloss': -1} 47 | monitor_mode: 'max' 48 | -------------------------------------------------------------------------------- /model_zoo/LorentzFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/LorentzFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .LorentzFM import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/MaskNet/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/MaskNet/README.md -------------------------------------------------------------------------------- /model_zoo/MaskNet/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/MaskNet/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | MaskNet_test: 16 | model: MaskNet 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | dnn_hidden_activations: relu 29 | model_type: SerialMaskNet 30 | parallel_num_blocks: 1 31 | parallel_block_dim: 64 32 | reduction_ratio: 1 33 | emb_layernorm: True 34 | net_layernorm: True 35 | net_dropout: 0 36 | epochs: 1 37 | shuffle: True 38 | seed: 2021 39 | monitor: 'AUC' 40 | monitor_mode: 'max' 41 | 42 | MaskNet_default: # This is a config template 43 | model: MaskNet 44 | dataset_id: TBD 45 | loss: 'binary_crossentropy' 46 | metrics: ['logloss', 'AUC'] 47 | task: binary_classification 48 | optimizer: adam 49 | learning_rate: 1.e-3 50 | embedding_regularizer: 0 51 | net_regularizer: 0 52 | batch_size: 4096 53 | embedding_dim: 40 54 | dnn_hidden_units: [400, 400, 400] 55 | dnn_hidden_activations: relu 56 | model_type: SerialMaskNet 57 | parallel_num_blocks: 1 58 | parallel_block_dim: 64 59 | reduction_ratio: 1 60 | emb_layernorm: True 61 | net_layernorm: True 62 | net_dropout: 0 63 | epochs: 100 64 | shuffle: True 65 | seed: 2021 66 | monitor: 'AUC' 67 | monitor_mode: 'max' 68 | -------------------------------------------------------------------------------- /model_zoo/MaskNet/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/MaskNet/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .MaskNet import * 2 | 3 | -------------------------------------------------------------------------------- /model_zoo/NFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/NFM/README.md -------------------------------------------------------------------------------- /model_zoo/NFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/NFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | NFM_test: 16 | model: NFM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | hidden_units: [64, 32] 28 | hidden_activations: relu 29 | net_dropout: 0 30 | batch_norm: False 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | NFM_default: # This is a config template 38 | model: NFM 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | learning_rate: 1.e-3 45 | embedding_regularizer: 0 46 | net_regularizer: 0 47 | batch_size: 10000 48 | embedding_dim: 40 49 | hidden_units: [500, 500, 500] 50 | hidden_activations: relu 51 | net_dropout: 0 52 | batch_norm: False 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | -------------------------------------------------------------------------------- /model_zoo/NFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/NFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .NFM import * 2 | 3 | -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_tf/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/ONN/ONN_tf/README.md -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_tf/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_tfrecord: 3 | data_root: ../../../data/ 4 | data_format: tfrecord 5 | train_data: ../../../data/tiny_tfrecord/train.tfrecord 6 | valid_data: ../../../data/tiny_tfrecord/valid.tfrecord 7 | test_data: ../../../data/tiny_tfrecord/test.tfrecord 8 | 9 | -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_tf/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | ONN_test: 16 | model: ONN 17 | dataset_id: tiny_tfrecord 18 | loss: binary_crossentropy 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | hidden_units: [64, 32] 26 | hidden_activations: relu 27 | net_dropout: 0 28 | batch_norm: False 29 | batch_size: 128 30 | embedding_dim: 4 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | ONN_default: # This is a config template 38 | model: ONN 39 | dataset_id: TBD 40 | loss: binary_crossentropy 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | learning_rate: 1.e-3 45 | embedding_regularizer: 0 46 | net_regularizer: 0 47 | hidden_units: [64, 64, 64] 48 | hidden_activations: relu 49 | net_dropout: 0 50 | batch_norm: False 51 | batch_size: 10000 52 | embedding_dim: 2 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' 58 | 59 | -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_tf/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.0.0" 4 | -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_tf/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .ONN import ONN 2 | -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_torch/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/ONN/ONN_torch/README.md -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_torch/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../../data/ 4 | data_format: npz 5 | train_data: ../../../data/tiny_npz/train.npz 6 | valid_data: ../../../data/tiny_npz/valid.npz 7 | test_data: ../../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_torch/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/ONN/ONN_torch/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .ONN import * 2 | from .ONNv2 import * 3 | -------------------------------------------------------------------------------- /model_zoo/PEPNet/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/PEPNet/README.md -------------------------------------------------------------------------------- /model_zoo/PEPNet/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/PEPNet/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | 14 | PPNet_test: 15 | model: PPNet 16 | dataset_id: tiny_npz 17 | loss: 'binary_crossentropy' 18 | metrics: ['logloss', 'AUC'] 19 | task: binary_classification 20 | optimizer: adam 21 | learning_rate: 1.e-3 22 | embedding_regularizer: 0 23 | net_regularizer: 0 24 | batch_size: 128 25 | embedding_dim: 4 26 | gate_emb_dim: 4 27 | gate_priors: ["userid", "adgroup_id", "pid"] 28 | gate_hidden_dim: 64 29 | hidden_units: [64, 64, 64] 30 | hidden_activations: ReLU 31 | net_dropout: 0 32 | batch_norm: False 33 | epochs: 1 34 | shuffle: True 35 | seed: 20222023 36 | monitor: 'AUC' 37 | monitor_mode: 'max' 38 | 39 | PPNet_default: # This is a config template 40 | model: PPNet 41 | dataset_id: TBD 42 | loss: 'binary_crossentropy' 43 | metrics: ['logloss', 'AUC'] 44 | task: binary_classification 45 | optimizer: adam 46 | learning_rate: 1.e-3 47 | embedding_regularizer: 0 48 | net_regularizer: 0 49 | batch_size: 10000 50 | embedding_dim: 16 51 | gate_emb_dim: 16 52 | gate_priors: ["userid", "adgroup_id", "pid"] 53 | gate_hidden_dim: 64 54 | hidden_units: [64, 64, 64] 55 | hidden_activations: ReLU 56 | net_dropout: 0 57 | batch_norm: False 58 | epochs: 100 59 | shuffle: True 60 | seed: 20222023 61 | monitor: 'AUC' 62 | monitor_mode: 'max' 63 | 64 | -------------------------------------------------------------------------------- /model_zoo/PEPNet/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/PEPNet/src/PEPNet.py: -------------------------------------------------------------------------------- 1 | # Todo -------------------------------------------------------------------------------- /model_zoo/PEPNet/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .PPNet import * 2 | from .PEPNet import * 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/PNN/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/PNN/README.md -------------------------------------------------------------------------------- /model_zoo/PNN/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/PNN/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | PNN_test: 16 | model: PNN 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | hidden_units: [64, 32] 28 | hidden_activations: relu 29 | net_dropout: 0 30 | batch_norm: False 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | PNN_default: # This is a config template 38 | model: PNN 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | hidden_units: [300, 300, 300] 45 | hidden_activations: relu 46 | embedding_regularizer: 0 47 | net_regularizer: 0 48 | learning_rate: 1.e-3 49 | net_dropout: 0 50 | batch_norm: False 51 | batch_size: 10000 52 | embedding_dim: 40 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' -------------------------------------------------------------------------------- /model_zoo/PNN/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/PNN/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .PNN import * 2 | 3 | -------------------------------------------------------------------------------- /model_zoo/SAM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/SAM/README.md -------------------------------------------------------------------------------- /model_zoo/SAM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/SAM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | SAM_test: 16 | model: SAM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | interaction_type: "SAM2E" 28 | aggregation: "concat" 29 | num_interaction_layers: 2 30 | use_residual: False 31 | net_dropout: 0 32 | epochs: 1 33 | shuffle: True 34 | seed: 2022 35 | monitor: 'AUC' 36 | monitor_mode: 'max' 37 | 38 | SAM_default: # This is a config template 39 | model: SAM 40 | dataset_id: TBD 41 | loss: 'binary_crossentropy' 42 | metrics: ['logloss', 'AUC'] 43 | task: binary_classification 44 | optimizer: adam 45 | learning_rate: 1.e-3 46 | embedding_regularizer: 0 47 | net_regularizer: 0 48 | batch_size: 4096 49 | embedding_dim: 40 50 | interaction_type: "SAM2E" 51 | aggregation: "concat" 52 | num_interaction_layers: 3 53 | use_residual: False 54 | net_dropout: 0 55 | epochs: 100 56 | shuffle: True 57 | seed: 2022 58 | monitor: 'AUC' 59 | monitor_mode: 'max' -------------------------------------------------------------------------------- /model_zoo/SAM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/SAM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .SAM import * 2 | 3 | -------------------------------------------------------------------------------- /model_zoo/TransAct/README.md: -------------------------------------------------------------------------------- 1 | # TransAct 2 | 3 | > Xue Xia, Pong Eksombatchai, Nikil Pancha, Dhruvil Deven Badani, Po-Wei Wang, Neng Gu, Saurabh Vishwas Joshi, Nazanin Farahpour, Zhiyuan Zhang, Andrew Zhai. [TransAct: Transformer-based Realtime User Action Model for Recommendation at Pinterest](https://arxiv.org/abs/2306.00248), in KDD 2023. 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/TransAct/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_seq: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_seq/train.npz 6 | valid_data: ../../data/tiny_seq/valid.npz 7 | test_data: ../../data/tiny_seq/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/TransAct/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/TransAct/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .TransAct import * 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_tf/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/WideDeep/WideDeep_tf/README.md -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_tf/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_tfrecord: 3 | data_root: ../../../data/ 4 | data_format: tfrecord 5 | train_data: ../../../data/tiny_tfrecord/train.tfrecord 6 | valid_data: ../../../data/tiny_tfrecord/valid.tfrecord 7 | test_data: ../../../data/tiny_tfrecord/test.tfrecord 8 | 9 | -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_tf/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | WideDeep_test: 16 | model: WideDeep 17 | dataset_id: tiny_tfrecord 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | wide_learning_rate: 1.e-4 22 | deep_learning_rate: 1.e-3 23 | deep_optimizer: adam 24 | hidden_units: [64, 32] 25 | hidden_activations: relu 26 | net_regularizer: 0 27 | embedding_regularizer: 0 28 | batch_norm: False 29 | net_dropout: 0 30 | batch_size: 8 31 | embedding_dim: 4 32 | epochs: 1 33 | shuffle: True 34 | seed: 2019 35 | monitor: 'AUC' 36 | monitor_mode: 'max' 37 | 38 | WideDeep_default: # This is a config template 39 | model: WideDeep 40 | dataset_id: TBD 41 | loss: 'binary_crossentropy' 42 | metrics: ['logloss', 'AUC'] 43 | task: binary_classification 44 | deep_optimizer: adam 45 | wide_learning_rate: 1.e-4 46 | deep_learning_rate: 1.e-3 47 | hidden_units: [300, 300, 300] 48 | hidden_activations: relu 49 | net_regularizer: 0 50 | embedding_regularizer: 0 51 | batch_norm: False 52 | net_dropout: 0 53 | batch_size: 10000 54 | embedding_dim: 40 55 | epochs: 100 56 | shuffle: True 57 | seed: 2019 58 | monitor: {'AUC': 1, 'logloss': -1} 59 | monitor_mode: 'max' 60 | 61 | 62 | -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_tf/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.0.0" 4 | -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_tf/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .WideDeep import WideDeep 2 | -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_torch/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/WideDeep/WideDeep_torch/README.md -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_torch/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../../data/ 4 | data_format: npz 5 | train_data: ../../../data/tiny_npz/train.npz 6 | valid_data: ../../../data/tiny_npz/valid.npz 7 | test_data: ../../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_torch/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | WideDeep_test: 16 | model: WideDeep 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | hidden_units: [64, 32] 24 | hidden_activations: relu 25 | net_regularizer: 0 26 | embedding_regularizer: 0 27 | batch_norm: False 28 | net_dropout: 0 29 | batch_size: 8 30 | embedding_dim: 4 31 | epochs: 1 32 | shuffle: True 33 | seed: 2019 34 | monitor: 'AUC' 35 | monitor_mode: 'max' 36 | 37 | WideDeep_default: # This is a config template 38 | model: WideDeep 39 | dataset_id: TBD 40 | loss: 'binary_crossentropy' 41 | metrics: ['logloss', 'AUC'] 42 | task: binary_classification 43 | optimizer: adam 44 | learning_rate: 1.e-3 45 | hidden_units: [300, 300, 300] 46 | hidden_activations: relu 47 | net_regularizer: 0 48 | embedding_regularizer: 0 49 | batch_norm: False 50 | net_dropout: 0 51 | batch_size: 10000 52 | embedding_dim: 40 53 | epochs: 100 54 | shuffle: True 55 | seed: 2019 56 | monitor: {'AUC': 1, 'logloss': -1} 57 | monitor_mode: 'max' -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_torch/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/WideDeep/WideDeep_torch/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .WideDeep import * 2 | 3 | -------------------------------------------------------------------------------- /model_zoo/WuKong/README.md: -------------------------------------------------------------------------------- 1 | # WuKong 2 | 3 | > Buyun Zhang, Liang Luo, Yuxin Chen, Jade Nie, Xi Liu, Daifeng Guo, Yanli Zhao, Shen Li, Yuchen Hao, Yantao Yao, Guna Lakshminarayanan, Ellie Dingqiao Wen, Jongsoo Park, Maxim Naumov, Wenlin Chen. [Wukong: Towards a Scaling Law for Large-Scale Recommendation](https://arxiv.org/abs/2403.02545), in Arxiv 2024. -------------------------------------------------------------------------------- /model_zoo/WuKong/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz -------------------------------------------------------------------------------- /model_zoo/WuKong/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | WuKong_default: # This is a config template 16 | model: WuKong 17 | dataset_id: TBD 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.0e-3 23 | embedding_regularizer: 0 24 | net_regularizer: 0 25 | batch_size: 10000 26 | embedding_dim: 64 27 | num_layers: 8 28 | compression_dim: 40 29 | fmb_units: [200,200] 30 | fmb_dim: 40 31 | project_dim: 8 32 | dropout_rate: 0.2 33 | hidden_activations: relu 34 | mlp_hidden_units: [32,32] 35 | epochs: 100 36 | shuffle: True 37 | seed: 2024 38 | monitor: {'AUC': 1, 'logloss': -1} 39 | monitor_mode: 'max' 40 | 41 | WuKong_test: 42 | model: WuKong 43 | dataset_id: tiny_npz 44 | loss: 'binary_crossentropy' 45 | metrics: ['logloss', 'AUC'] 46 | task: binary_classification 47 | optimizer: adam 48 | learning_rate: 1.0e-3 49 | embedding_regularizer: 0 50 | net_regularizer: 0 51 | batch_size: 2048 52 | embedding_dim: 64 53 | num_layers: 4 54 | compression_dim: 32 55 | fmb_units: [128,128,128] 56 | fmb_dim: 32 57 | project_dim: 24 58 | dropout_rate: 0.2 59 | hidden_activations: relu 60 | mlp_hidden_units: [64] 61 | epochs: 1 62 | shuffle: True 63 | seed: 2024 64 | monitor: 'AUC' 65 | monitor_mode: 'max' 66 | -------------------------------------------------------------------------------- /model_zoo/WuKong/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/WuKong/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .WuKong import * -------------------------------------------------------------------------------- /model_zoo/__init__.py: -------------------------------------------------------------------------------- 1 | from .AFM.src import AFM 2 | from .AFN.src import AFN 3 | from .AOANet.src import AOANet 4 | from .AutoInt.src import AutoInt 5 | from .BST.src import BST 6 | from .CCPM.src import CCPM 7 | from .DCN.DCN_torch.src import DCN 8 | from .DCNv2.src import DCNv2 9 | from .DeepCrossing.src import DeepCrossing 10 | from .DeepFM.DeepFM_torch.src import DeepFM 11 | from .DeepIM.src import DeepIM 12 | from .DESTINE.src import DESTINE 13 | from .DIEN.src import DIEN 14 | from .DIN.src import DIN 15 | from .DLRM.src import DLRM 16 | from .DMIN.src import DMIN 17 | from .DMR.src import DMR 18 | from .DNN.DNN_torch.src import DNN 19 | from .DSSM.src import DSSM 20 | from .EDCN.src import EDCN 21 | from .FFM.src import FFM, FFMv2 22 | from .FGCNN.src import FGCNN 23 | from .FiBiNET.src import FiBiNET 24 | from .FiGNN.src import FiGNN 25 | from .FinalMLP.src import FinalMLP 26 | from .FinalNet.src import FinalNet 27 | from .FLEN.src import FLEN 28 | from .FM.src import FM 29 | from .FmFM.src import FmFM 30 | from .FwFM.src import FwFM 31 | from .HFM.src import HFM 32 | from .HOFM.src import HOFM 33 | from .InterHAt.src import InterHAt 34 | from .LorentzFM.src import LorentzFM 35 | from .LR.src import LR 36 | from .MaskNet.src import MaskNet 37 | from .NFM.src import NFM 38 | from .ONN.ONN_torch.src import ONN, ONNv2 39 | from .PNN.src import PNN 40 | from .SAM.src import SAM 41 | from .WideDeep.WideDeep_torch.src import WideDeep 42 | from .xDeepFM.src import xDeepFM 43 | from .PEPNet.src import PPNet 44 | from .TransAct.src import TransAct 45 | from .multitask import ShareBottom, MMoE, PLE 46 | from .EulerNet.src import EulerNet 47 | from .WuKong.src import WuKong 48 | from .GDCN.src import GDCN 49 | -------------------------------------------------------------------------------- /model_zoo/multitask/MMoE/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | tiny_mtl: 2 | data_root: ../../../data/ 3 | data_format: csv 4 | train_data: ../../../data/tiny_mtl/train.csv 5 | valid_data: ../../../data/tiny_mtl/test.csv 6 | test_data: ../../../data/tiny_mtl/test.csv 7 | min_categr_count: 1 8 | feature_cols: 9 | [ { name: [ categorical_1,categorical_2,categorical_3,categorical_4,categorical_5,categorical_6, 10 | categorical_7,categorical_8,categorical_9,categorical_10,categorical_11,categorical_12, 11 | categorical_13,categorical_14,categorical_15,categorical_16 ], 12 | active: True, dtype: str, type: categorical }, 13 | { name: [ numerical_1,numerical_2,numerical_3,numerical_4,numerical_5,numerical_6,numerical_7,numerical_8, 14 | numerical_9,numerical_10,numerical_11,numerical_12,numerical_13,numerical_14,numerical_15,numerical_16, 15 | numerical_17,numerical_18,numerical_19,numerical_20,numerical_21,numerical_22,numerical_23,numerical_24, 16 | numerical_25,numerical_26,numerical_27,numerical_28,numerical_29,numerical_30,numerical_31,numerical_32, 17 | numerical_33,numerical_34,numerical_35,numerical_36,numerical_37,numerical_38,numerical_39,numerical_40, 18 | numerical_41,numerical_42,numerical_43,numerical_44,numerical_45,numerical_46,numerical_47,numerical_48, 19 | numerical_49,numerical_50,numerical_51,numerical_52,numerical_53,numerical_54,numerical_55,numerical_56, 20 | numerical_57,numerical_58,numerical_59,numerical_60,numerical_61,numerical_62,numerical_63 ], 21 | active: True, dtype: float, type: numeric } ] 22 | label_col: [ { name: click, dtype: float }, 23 | { name: conversion, dtype: float } ] -------------------------------------------------------------------------------- /model_zoo/multitask/MMoE/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/multitask/MMoE/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .MMoE import * -------------------------------------------------------------------------------- /model_zoo/multitask/PLE/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | tiny_mtl: 2 | data_root: ../../../data/ 3 | data_format: csv 4 | train_data: ../../../data/tiny_mtl/train.csv 5 | valid_data: ../../../data/tiny_mtl/test.csv 6 | test_data: ../../../data/tiny_mtl/test.csv 7 | min_categr_count: 1 8 | feature_cols: 9 | [ { name: [ categorical_1,categorical_2,categorical_3,categorical_4,categorical_5,categorical_6, 10 | categorical_7,categorical_8,categorical_9,categorical_10,categorical_11,categorical_12, 11 | categorical_13,categorical_14,categorical_15,categorical_16 ], 12 | active: True, dtype: str, type: categorical }, 13 | { name: [ numerical_1,numerical_2,numerical_3,numerical_4,numerical_5,numerical_6,numerical_7,numerical_8, 14 | numerical_9,numerical_10,numerical_11,numerical_12,numerical_13,numerical_14,numerical_15,numerical_16, 15 | numerical_17,numerical_18,numerical_19,numerical_20,numerical_21,numerical_22,numerical_23,numerical_24, 16 | numerical_25,numerical_26,numerical_27,numerical_28,numerical_29,numerical_30,numerical_31,numerical_32, 17 | numerical_33,numerical_34,numerical_35,numerical_36,numerical_37,numerical_38,numerical_39,numerical_40, 18 | numerical_41,numerical_42,numerical_43,numerical_44,numerical_45,numerical_46,numerical_47,numerical_48, 19 | numerical_49,numerical_50,numerical_51,numerical_52,numerical_53,numerical_54,numerical_55,numerical_56, 20 | numerical_57,numerical_58,numerical_59,numerical_60,numerical_61,numerical_62,numerical_63 ], 21 | active: True, dtype: float, type: numeric } ] 22 | label_col: [ { name: click, dtype: float }, 23 | { name: conversion, dtype: float } ] -------------------------------------------------------------------------------- /model_zoo/multitask/PLE/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/multitask/PLE/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .PLE import * -------------------------------------------------------------------------------- /model_zoo/multitask/ShareBottom/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | tiny_mtl: 2 | data_root: ../../../data/ 3 | data_format: csv 4 | train_data: ../../../data/tiny_mtl/train.csv 5 | valid_data: ../../../data/tiny_mtl/test.csv 6 | test_data: ../../../data/tiny_mtl/test.csv 7 | min_categr_count: 1 8 | feature_cols: 9 | [ { name: [ categorical_1,categorical_2,categorical_3,categorical_4,categorical_5,categorical_6, 10 | categorical_7,categorical_8,categorical_9,categorical_10,categorical_11,categorical_12, 11 | categorical_13,categorical_14,categorical_15,categorical_16 ], 12 | active: True, dtype: str, type: categorical }, 13 | { name: [ numerical_1,numerical_2,numerical_3,numerical_4,numerical_5,numerical_6,numerical_7,numerical_8, 14 | numerical_9,numerical_10,numerical_11,numerical_12,numerical_13,numerical_14,numerical_15,numerical_16, 15 | numerical_17,numerical_18,numerical_19,numerical_20,numerical_21,numerical_22,numerical_23,numerical_24, 16 | numerical_25,numerical_26,numerical_27,numerical_28,numerical_29,numerical_30,numerical_31,numerical_32, 17 | numerical_33,numerical_34,numerical_35,numerical_36,numerical_37,numerical_38,numerical_39,numerical_40, 18 | numerical_41,numerical_42,numerical_43,numerical_44,numerical_45,numerical_46,numerical_47,numerical_48, 19 | numerical_49,numerical_50,numerical_51,numerical_52,numerical_53,numerical_54,numerical_55,numerical_56, 20 | numerical_57,numerical_58,numerical_59,numerical_60,numerical_61,numerical_62,numerical_63 ], 21 | active: True, dtype: float, type: numeric } ] 22 | label_col: [ { name: click, dtype: float }, 23 | { name: conversion, dtype: float } ] -------------------------------------------------------------------------------- /model_zoo/multitask/ShareBottom/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | ### Base: This base setting will be inherited by all the expid configs. 2 | Base: 3 | model_root: './checkpoints/' 4 | num_workers: 3 5 | verbose: 1 6 | early_stop_patience: 2 7 | pickle_feature_encoder: True 8 | save_best_only: True 9 | eval_steps: null 10 | debug_mode: False 11 | group_id: null 12 | use_features: null 13 | feature_specs: null 14 | feature_config: null 15 | 16 | ### ModelName_default: This is a config template for hyper-tuning use 17 | ShareBottom_default: 18 | model: ShareBottom 19 | dataset_id: TBD 20 | loss: ['binary_crossentropy', 'binary_crossentropy'] 21 | metrics: ['logloss', 'AUC'] 22 | task: ['binary_classification', 'binary_classification'] 23 | num_tasks: 2 24 | optimizer: adam 25 | learning_rate: 1.e-3 26 | bottom_hidden_units: [512, 256, 128] 27 | tower_hidden_units: [128, 64] 28 | hidden_activations: relu 29 | net_regularizer: 0 30 | embedding_regularizer: 1.e-6 31 | batch_norm: False 32 | net_dropout: 0 33 | batch_size: 128 34 | embedding_dim: 64 35 | epochs: 100 36 | shuffle: True 37 | seed: 2023 38 | monitor: 'AUC' 39 | monitor_mode: 'max' 40 | 41 | ### ModelName_test: This is a config for test only 42 | ShareBottom_test: 43 | model: ShareBottom 44 | dataset_id: tiny_mtl 45 | loss: ['binary_crossentropy', 'binary_crossentropy'] 46 | metrics: ['logloss', 'AUC'] 47 | task: ['binary_classification', 'binary_classification'] 48 | num_tasks: 2 49 | optimizer: adam 50 | learning_rate: 1.e-3 51 | bottom_hidden_units: [512, 256, 128] 52 | tower_hidden_units: [128, 64] 53 | hidden_activations: relu 54 | net_regularizer: 0 55 | embedding_regularizer: 1.e-6 56 | batch_norm: False 57 | net_dropout: 0 58 | batch_size: 128 59 | embedding_dim: 64 60 | epochs: 1 61 | shuffle: True 62 | seed: 2023 63 | monitor: 'AUC' 64 | monitor_mode: 'max' 65 | -------------------------------------------------------------------------------- /model_zoo/multitask/ShareBottom/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/multitask/ShareBottom/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .ShareBottom import ShareBottom 2 | -------------------------------------------------------------------------------- /model_zoo/multitask/__init__.py: -------------------------------------------------------------------------------- 1 | from .MMoE.src import MMoE 2 | from .ShareBottom.src import ShareBottom 3 | from .PLE.src import PLE 4 | -------------------------------------------------------------------------------- /model_zoo/xDeepFM/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/reczoo/FuxiCTR/5c0c3084f93104f04f1f6186829ae00621eb0e35/model_zoo/xDeepFM/README.md -------------------------------------------------------------------------------- /model_zoo/xDeepFM/config/dataset_config.yaml: -------------------------------------------------------------------------------- 1 | ### Tiny data for tests only 2 | tiny_npz: 3 | data_root: ../../data/ 4 | data_format: npz 5 | train_data: ../../data/tiny_npz/train.npz 6 | valid_data: ../../data/tiny_npz/valid.npz 7 | test_data: ../../data/tiny_npz/test.npz 8 | -------------------------------------------------------------------------------- /model_zoo/xDeepFM/config/model_config.yaml: -------------------------------------------------------------------------------- 1 | Base: 2 | model_root: './checkpoints/' 3 | num_workers: 3 4 | verbose: 1 5 | early_stop_patience: 2 6 | pickle_feature_encoder: True 7 | save_best_only: True 8 | eval_steps: null 9 | debug_mode: False 10 | group_id: null 11 | use_features: null 12 | feature_specs: null 13 | feature_config: null 14 | 15 | xDeepFM_test: 16 | model: xDeepFM 17 | dataset_id: tiny_npz 18 | loss: 'binary_crossentropy' 19 | metrics: ['logloss', 'AUC'] 20 | task: binary_classification 21 | optimizer: adam 22 | learning_rate: 1.e-3 23 | embedding_regularizer: 1.e-8 24 | net_regularizer: 0 25 | batch_size: 128 26 | embedding_dim: 4 27 | dnn_hidden_units: [64, 32] 28 | cin_layer_units: [10, 10] 29 | hidden_activations: relu 30 | net_dropout: 0 31 | batch_norm: False 32 | epochs: 1 33 | shuffle: True 34 | seed: 2019 35 | monitor: 'AUC' 36 | monitor_mode: 'max' 37 | 38 | xDeepFM_default: # This is a config template 39 | model: xDeepFM 40 | dataset_id: TBD 41 | loss: 'binary_crossentropy' 42 | metrics: ['logloss', 'AUC'] 43 | task: binary_classification 44 | optimizer: adam 45 | learning_rate: 1.0e-3 46 | embedding_regularizer: 0 47 | net_regularizer: 0 48 | batch_size: 10000 49 | embedding_dim: 40 50 | dnn_hidden_units: [500, 500, 500] 51 | cin_hidden_units: [32, 32, 32] 52 | hidden_activations: relu 53 | net_dropout: 0 54 | batch_norm: False 55 | epochs: 100 56 | shuffle: True 57 | seed: 2019 58 | monitor: {'AUC': 1, 'logloss': -1} 59 | monitor_mode: 'max' -------------------------------------------------------------------------------- /model_zoo/xDeepFM/fuxictr_version.py: -------------------------------------------------------------------------------- 1 | # pip install -U fuxictr 2 | import fuxictr 3 | assert fuxictr.__version__ >= "2.3.7" 4 | -------------------------------------------------------------------------------- /model_zoo/xDeepFM/src/__init__.py: -------------------------------------------------------------------------------- 1 | from .xDeepFM import * 2 | 3 | -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | keras_preprocessing 2 | pandas 3 | PyYAML 4 | scikit-learn 5 | numpy 6 | h5py 7 | tqdm 8 | pyarrow 9 | polars<=1.0.0 10 | -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- 1 | import setuptools 2 | 3 | with open("README.md", "r", encoding="utf-8") as fd: 4 | long_description = fd.read() 5 | 6 | setuptools.setup( 7 | name="fuxictr", 8 | version="2.3.8", 9 | author="RECZOO", 10 | author_email="reczoo@users.noreply.github.com", 11 | description="A configurable, tunable, and reproducible library for CTR prediction", 12 | long_description=long_description, 13 | long_description_content_type="text/markdown", 14 | url="https://github.com/reczoo/FuxiCTR", 15 | download_url='https://github.com/reczoo/FuxiCTR/tags', 16 | packages=setuptools.find_packages( 17 | exclude=["model_zoo", "tests", "data", "docs", "demo"]), 18 | include_package_data=True, 19 | python_requires=">=3.6", 20 | install_requires=["keras_preprocessing", "pandas", "PyYAML>=5.1", "scikit-learn", 21 | "numpy", "h5py", "tqdm", "pyarrow", "polars"], 22 | classifiers=( 23 | "License :: OSI Approved :: Apache Software License", 24 | "Operating System :: OS Independent", 25 | 'Intended Audience :: Developers', 26 | 'Intended Audience :: Education', 27 | 'Intended Audience :: Science/Research', 28 | 'Programming Language :: Python :: 3', 29 | 'Topic :: Scientific/Engineering', 30 | 'Topic :: Scientific/Engineering :: Artificial Intelligence', 31 | 'Topic :: Software Development', 32 | 'Topic :: Software Development :: Libraries', 33 | 'Topic :: Software Development :: Libraries :: Python Modules', 34 | ), 35 | license="Apache-2.0 License", 36 | keywords=['ctr prediction', 'recommender systems', 37 | 'ctr', 'cvr', 'pytorch'], 38 | ) 39 | -------------------------------------------------------------------------------- /tests/test_demo.sh: -------------------------------------------------------------------------------- 1 | #! /bin/sh 2 | cd "$(pwd)/../demo" 3 | 4 | echo "=== Testing example1 ===" && python example1_build_dataset_to_parquet.py && \ 5 | echo "=== Testing example2 ===" && python example2_DeepFM_with_parquet_input.py && \ 6 | echo "=== Testing example3 ===" && python example3_DeepFM_with_npz_input.py && \ 7 | echo "=== Testing example4 ===" && python example4_DeepFM_with_csv_input.py && \ 8 | echo "=== Testing example5 ===" && python example5_DeepFM_with_pretrained_emb.py && \ 9 | echo "=== Testing example6 ===" && python example6_DIN_with_sequence_feature.py && \ 10 | echo "=== Testing example7 ===" && python example7_DeepFM_with_customized_preprocess.py && \ 11 | 12 | echo "All tests done." 13 | -------------------------------------------------------------------------------- /tests/test_tensorflow.sh: -------------------------------------------------------------------------------- 1 | #! /bin/sh 2 | home="$(pwd)/../model_zoo" 3 | 4 | echo "=== Testing DNN_tf ===" && cd $home/DNN/DNN_tf && python run_expid.py --expid DNN_test && \ 5 | echo "=== Testing DCN_tf ===" && cd $home/DCN/DCN_tf && python run_expid.py --expid DCN_test && \ 6 | echo "=== Testing WideDeep_tf ===" && cd $home/WideDeep/WideDeep_tf && python run_expid.py --expid WideDeep_test && \ 7 | 8 | echo "All tests done." 9 | --------------------------------------------------------------------------------