FuxiCTR

This repo is the community dev version of the official release at huawei-noah/benchmark/FuxiCTR.

Click-through rate (CTR) prediction is an critical task for many industrial applications such as online advertising, recommender systems, and sponsored search. FuxiCTR provides an open-source library for CTR prediction, with key features in configurability, tunability, and reproducibility. It also supports the building of the BARS-CTR-Prediction benchmark, which aims for open benchmarking for CTR prediction.

👉 If you find our code or benchmarks helpful in your research, please kindly cite the following paper.

Jieming Zhu, Jinyang Liu, Shuai Yang, Qi Zhang, Xiuqiang He. Open Benchmarking for Click-Through Rate Prediction. The 30th ACM International Conference on Information and Knowledge Management (CIKM), 2021.

Model List

Installation

Please follow the guide for installation. In particular, FuxiCTR has the following dependent requirements.

• python 3.6
• pytorch v1.0/v1.1
• pyyaml >=5.1
• scikit-learn
• pandas
• numpy
• h5py
• tqdm

Get Started

1. Run the demo to understand the overall workflow

2. Run a model with dataset and model config files

3. Preprocess raw csv data to h5 data

4. Run a model with h5 data as input

5. How to make configurations?

6. Tune the model hyper-parameters via grid search

7. Run a model with sequence features

8. Run a model with pretrained embeddings

Code Structure

Check an overview of code structure for more details on API design.

Open Benchmarking

If you are looking for the benchmarking settings and results on the state-of-the-art CTR prediction models, please refer to the BARS-CTR-Prediction benchmark. By clicking on the "SOTA Results", you will find the benchmarking results along with the corresponding reproducing steps.

Discussion

Welcome to join our WeChat group for any questions and discussions.

Join Us

We have open positions for internships and full-time jobs. If you are interested in research and practice in recommender systems, please send your CV to [email protected].