Welcome to DeepCTR’s documentation!¶
DeepCTR is a Easy-to-use , Modular and Extendible package of deep-learning based CTR models along with lots of core components layer which can be used to easily build custom models.You can use any complex model with model.fit() and model.predict().
Provide
tf.keras.Modellike interface for quick experiment. exampleProvide
tensorflow estimatorinterface for large scale data and distributed training. exampleIt is compatible with both
tf 1.15andtf 2.x.
Install a TensorFlow build that matches your Python, NumPy, CPU/GPU, and operating system first, then install DeepCTR. Use public tensorflow.keras APIs in your own code and avoid mixing them with private tensorflow.python.keras imports.
Let’s Get Started! (Chinese Introduction)
You can read the latest code and related projects
DeepMatch: https://github.com/shenweichen/DeepMatch
DeepCTR-Torch: https://github.com/shenweichen/DeepCTR-Torch
News¶
04/16/2026 : Support higher tensorflow version . Changelog
11/10/2022 : Add EDCN . Changelog
10/15/2022 : Support python 3.9 , 3.10 . Changelog
DisscussionGroup¶
公众号：浅梦学习笔记 wechat ID: deepctrbot
Home:
- Quick-Start
- Features
- Examples
- Classification: Criteo
- Classification: Criteo with feature hashing on the fly
- Regression: Movielens
- Multi-value Input : Movielens
- Multi-value Input : Movielens with feature hashing on the fly
- Hash Layer with pre-defined key-value vocabulary
- Estimator with TFRecord: Classification Criteo
- Estimator with Pandas DataFrame: Classification Criteo
- MultiTask Learning:MMOE
- FAQ
- 1. Save or load weights/models
- 2. Set learning rate and use earlystopping
- 3. Get the attentional weights of feature interactions in AFM
- 4. How to extract the embedding vectors in deepfm?
- 5. How to add a long dense feature vector as a input to the model?
- 6. How to use pretrained weights to initialize embedding weights and frozen embedding weights?
- 7. How to run the demo with GPU ?
- 8. How to avoid TensorFlow, Keras, h5py, or NumPy compatibility errors?
- 9. How to run the demo with multiple GPUs
- History
Cookbook:
- Sequence Feature Cookbook
- One Row Means One Prediction Target
- Feature Column Shapes
- Padding and Length
- Multi-Value Feature or Multi-Hot Vector
- Multiple VarLenSparseFeat Fields
- Sharing Embeddings
- DIN and BST History Feature Names
- DIN Does Not Support VarLen of VarLen
- Dense History Features in DIN
- DSIN Input Format
- DSIN Labels and Sessions
- Estimator and TFRecord Inputs
- Quick Checklist