From fab38a46e84672463a9630cad92a322c52d0fe88 Mon Sep 17 00:00:00 2001 From: yaoxuefeng Date: Sat, 9 May 2020 11:24:09 +0800 Subject: [PATCH] modify rank readme --- models/rank/readme.md | 46 +++++++++++++++++++++---------------------- 1 file changed, 23 insertions(+), 23 deletions(-) diff --git a/models/rank/readme.md b/models/rank/readme.md index ff460a46..614a34dc 100644 --- a/models/rank/readme.md +++ b/models/rank/readme.md @@ -1,7 +1,7 @@ -# Rank模型库 +# 排序模型库 ## 简介 -我们提供了常见的ctr任务中使用的模型,包括 [dnn](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/dnn)、[dcn](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/dcn)、[deepfm](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/deepfm)、 [xdeepfm](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/xdeepfm)、[din](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/din)、[wide&deep](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/wide_deep)。 +我们提供了常见的排序任务中使用的模型算法,包括 [多层神经网络](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/dnn)、[Deep Cross Network](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/dcn)、[DeepFM](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/deepfm)、 [xDeepFM](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/xdeepfm)、[Deep Interest Network](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/din)、[Wide&Deep](http://gitlab.baidu.com/tangwei12/paddlerec/tree/develop/models/rank/wide_deep)。 模型算法库在持续添加中,欢迎关注。 @@ -23,11 +23,11 @@ | 模型 | 简介 | 论文 | | :------------------: | :--------------------: | :---------: | | DNN | 多层神经网络 | -- | -| wide&deep | Deep + wide(LR) | [论文链接](https://dl.acm.org/doi/abs/10.1145/2988450.2988454)(2016) | -| DeepFM | Deep + FM 并行 | [论文链接](https://arxiv.org/abs/1703.04247)(2017) | -| xDeepFM | DeepFM升级版 | [论文链接](https://dl.acm.org/doi/abs/10.1145/3219819.3220023)(2018) | -| DCN | wide升级为Cross Layer Network | [论文链接](https://dl.acm.org/doi/abs/10.1145/3124749.3124754)(2017) | -| DIN | Embeddding层引入attention机制 | [论文链接](https://dl.acm.org/doi/abs/10.1145/3219819.3219823)(2018) | +| wide&deep | Deep + wide(LR) | [Wide & Deep Learning for Recommender Systems](https://dl.acm.org/doi/abs/10.1145/2988450.2988454)(2016) | +| DeepFM | Deep + FM 并行 | [DeepFM: A Factorization-Machine based Neural Network for CTR Prediction](https://arxiv.org/abs/1703.04247)(2017) | +| xDeepFM | DeepFM升级版 | [xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems](https://dl.acm.org/doi/abs/10.1145/3219819.3220023)(2018) | +| DCN | wide升级为Cross Layer Network | [Deep & Cross Network for Ad Click Predictions](https://dl.acm.org/doi/abs/10.1145/3124749.3124754)(2017) | +| DIN | Embeddding层引入attention机制 | [Deep Interest Network for Click-Through Rate Prediction](https://dl.acm.org/doi/abs/10.1145/3219819.3219823)(2018) | ## 使用教程 ### 数据处理 @@ -37,22 +37,22 @@ ## 效果对比 ### 模型效果列表 -| 数据集 | 模型 | 单机测试集指标 | 详情 | -| :------------------: | :--------------------: | :---------: |:---------: | -| Criteo | DNN | auc:0.79395 | [更多](https://github.com/PaddlePaddle/models/tree/develop/PaddleRec/ctr/dnn#benchmark) | -| Criteo | DeepFM | logloss: 0.44797,
auc:0.8046 | [更多](https://github.com/PaddlePaddle/models/tree/develop/PaddleRec/ctr/deepfm#result) | -| Criteo | DCN | logloss: 0.44703564,
auc: 0.80654419 | [更多](https://github.com/PaddlePaddle/models/tree/develop/PaddleRec/ctr/dcn#%E7%BB%93%E6%9E%9C) | -| Demo数据集 | xDeepFM | acc: 0.48657,
auc:0.7308 | [更多](https://github.com/PaddlePaddle/models/tree/develop/PaddleRec/ctr/xdeepfm#%E5%8D%95%E6%9C%BA%E7%BB%93%E6%9E%9C) | -| Census-income Data | Wide&Deep | mean_acc:0.76195,
mean_auc:0.90577 | [更多](https://github.com/PaddlePaddle/models/tree/develop/PaddleRec/ctr/wide_deep#%E6%A8%A1%E5%9E%8B%E6%95%88%E6%9E%9C) | -| Amazon Product | DIN | logloss: 0.47005194,
auc: 0.863794952818 | [更多](https://github.com/PaddlePaddle/models/tree/develop/PaddleRec/ctr/din#%E9%A2%84%E6%B5%8B%E7%BB%93%E6%9E%9C%E7%A4%BA%E4%BE%8B) | +| 数据集 | 模型 | loss | 测试auc | acc | mae | +| :------------------: | :--------------------: | :---------: |:---------: | :---------: |:---------: | +| Criteo | DNN | -- | 0.79395 | -- | -- | +| Criteo | DeepFM | 0.44797 | 0.8046 | -- | -- | +| Criteo | DCN | 0.44703564 | 0.80654419 | -- | -- | +| Criteo | xDeepFM | -- | 0.7308 | 0.48657 | -- | +| Census-income Data | Wide&Deep | 0.76195(mean) | 0.90577(mean) | -- | -- | +| Amazon Product | DIN | 0.47005194 | 0.863794952818 | -- | -- | ## 分布式 ### 模型性能列表 -| 数据集 | 模型 | 单机 | 多机(同步) | 多机(异步) | -| :------------------: | :--------------------: | :---------: |:---------: |:---------: | -| Criteo | DNN | -- | -- | -- | -| Criteo | DeepFM | -- | -- | -- | -| Criteo | DCN | -- | -- | -- | -| Demo数据集 | xDeepFM | -- | -- | -- | -| Census-income Data | Wide&Deep | -- | -- | -- | -| Amazon Product | DIN | -- | -- | -- | +| 数据集 | 模型 | 单机 | 多机(同步) | 多机(异步) | GPU | +| :------------------: | :--------------------: | :---------: |:---------: |:---------: |:---------: | +| Criteo | DNN | -- | -- | -- | -- | +| Criteo | DeepFM | -- | -- | -- | -- | +| Criteo | DCN | -- | -- | -- | -- | +| Criteo | xDeepFM | -- | -- | -- | -- | +| Census-income Data | Wide&Deep | -- | -- | -- | -- | +| Amazon Product | DIN | -- | -- | -- | -- | -- GitLab