Home¶
Built on PyTorch, RecLib makes it easy to design and evaluate new deep learning models for recommender system, along with the infrastructure to easily run them in the cloud or on your laptop. RecLib was designed with the following principles:
Hyper-modular and lightweight. Use the parts which you like seamlessly with PyTorch.
Extensively tested and easy to extend. Test coverage is above 90% and the example models provide a template for contributions.
Take object oriented design seriously, making it easy to implement correct models without the pain.
Experiment friendly. Run reproducible experiments with as little as work possible.
RecLib includes reference implementations of high quality models for CTR, ad ranking and more (see https://github.com/tingkai-zhang/reclib#models).
RecLib is built and maintained by the Tingkai Zhang. The RecLib project is uniquely positioned to provide state of the art models with high quality engineering.
Package Reference
- reclib.common package
- reclib.data package
- reclib.models package
- Submodules
- reclib.models.afi module
- reclib.models.afm module
- reclib.models.dcn module
- reclib.models.dfm module
- reclib.models.ffm module
- reclib.models.fm module
- reclib.models.fnfm module
- reclib.models.fnn module
- reclib.models.lr module
- reclib.models.model module
- reclib.models.nfm module
- reclib.models.pnn module
- reclib.models.wd module
- reclib.models.xdeepfm module
- Module contents
- reclib.modules package
- reclib.nn package