diff --git a/models/rank/fm/config.yaml b/models/rank/fm/config.yaml index 8b72ab5d8ce62cf31be7eef631cfab03d3a6a7d3..6d4d96bfe5c207f68efdf5bbba9f2c03548d1d26 100644 --- a/models/rank/fm/config.yaml +++ b/models/rank/fm/config.yaml @@ -35,7 +35,7 @@ hyper_parameters: # 用户自定义配置 optimizer: class: Adam - learning_rate: 0.0001 + learning_rate: 0.001 sparse_feature_number: 1086460 sparse_feature_dim: 9 is_sparse: False @@ -48,7 +48,7 @@ mode: [train_runner,infer_runner] runner: - name: train_runner class: train - epochs: 1 + epochs: 2 device: cpu init_model_path: "" save_checkpoint_interval: 1 diff --git a/models/rank/fm/picture/1.jpg b/models/rank/fm/picture/1.jpg new file mode 100644 index 0000000000000000000000000000000000000000..09340a5e3e10abb9ef43926187048b48694f9b45 Binary files /dev/null and b/models/rank/fm/picture/1.jpg differ diff --git a/models/rank/fm/picture/2.jpg b/models/rank/fm/picture/2.jpg new file mode 100644 index 0000000000000000000000000000000000000000..2096e6a39bc96f72a903c7c0da85670323472452 Binary files /dev/null and b/models/rank/fm/picture/2.jpg differ diff --git a/models/rank/fm/picture/3.jpg b/models/rank/fm/picture/3.jpg new file mode 100644 index 0000000000000000000000000000000000000000..162efdb98f65ab421d9366b03bd6e5588b770717 Binary files /dev/null and b/models/rank/fm/picture/3.jpg differ diff --git a/models/rank/fm/picture/4.jpg b/models/rank/fm/picture/4.jpg new file mode 100644 index 0000000000000000000000000000000000000000..2c4273eb9ca5d27d29bc43f06e5cd98261fc036a Binary files /dev/null and b/models/rank/fm/picture/4.jpg differ diff --git a/models/rank/fm/readme.md b/models/rank/fm/readme.md index 6dd491eea87bbdab7abc97116402fe2321d7b800..9081d811b5d35068f650d231756c98156c8d494d 100644 --- a/models/rank/fm/readme.md +++ b/models/rank/fm/readme.md @@ -1,6 +1,37 @@ # 基于FM模型的点击率预估模型 -## 介绍 +以下是本例的简要目录结构及说明: + +``` +├── sample_data #样例数据 + ├── train + ├── sample_train.txt #训练数据样例 + ├── preprocess.py #数据处理程序 + ├── run.sh #数据一键处理脚本 + ├── download_preprocess.py #数据下载脚本 + ├── get_slot_data.py #格式整理程序 +├── __init__.py +├── README.md #文档 +├── model.py #模型文件 +├── config.yaml #配置文件 +``` + +注:在阅读该示例前,建议您先了解以下内容: + +[paddlerec入门教程](https://github.com/PaddlePaddle/PaddleRec/blob/master/README.md) + +## 内容 + +- [模型简介](#模型简介) +- [数据准备](#数据准备) +- [运行环境](#运行环境) +- [快速开始](#快速开始) +- [模型组网](#模型组网) +- [效果复现](#效果复现) +- [进阶使用](#进阶使用) +- [FAQ](#FAQ) + +## 模型简介 `CTR(Click Through Rate)`,即点击率,是“推荐系统/计算广告”等领域的重要指标,对其进行预估是商品推送/广告投放等决策的基础。简单来说,CTR预估对每次广告的点击情况做出预测,预测用户是点击还是不点击。CTR预估模型综合考虑各种因素、特征,在大量历史数据上训练,最终对商业决策提供帮助。本模型实现了下述论文中的FM模型: ```text @@ -20,244 +51,117 @@ ```bash