DeepCTR-Torch and DeepCTR
These are ecosystem siblings, representing two different implementations of the same DeepCTR framework: one in PyTorch (DeepCTR-Torch, actively maintained with many downloads) and an older, more general implementation (DeepCTR, with more stars but no recent downloads, suggesting it may be less actively maintained or superseded by the PyTorch version).
About DeepCTR-Torch
shenweichen/DeepCTR-Torch
【PyTorch】Easy-to-use,Modular and Extendible package of deep-learning based CTR models.
About DeepCTR
shenweichen/DeepCTR
Easy-to-use,Modular and Extendible package of deep-learning based CTR models .
# Technical Summary for DeepCTR Implements 20+ state-of-the-art CTR prediction architectures (DeepFM, xDeepFM, DCN, AutoInt, etc.) as composable tf.keras models with reusable embedding and attention layers. Supports both rapid prototyping via standard Keras APIs and production-scale distributed training through TensorFlow Estimators with TFRecord input pipelines, maintaining compatibility across TensorFlow 1.x and 2.x.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work