/ETRec

paper：Towards Efficient and Effective Transformers for Sequential Recommendation

Primary LanguagePythonMIT LicenseMIT

The Library of Efficient Transformers for Sequential Recommendation

Code for DASFAA-2023 submission:

Towards Efficient and Effective Transformers for Sequential Recommendation
(Running Title: Towards Efficient Transformers for Sequential Recommendation)

Note: this library is being updated continuously.

The library is built upon PyTorch and RecBole for reproducing recommendation algorithms based on Transformers and then exploring their effectiveness and efficiency.

Requirements

pytorch>=1.7.0
python>=3.7.0
recbole>=1.0.0

Implemented Models

The implemented models can be seen in the library of efficient Transformers.

path: /recbole/model/efficient_transformer_recommender/*
/recbole/model/transformer_layers.py
/recbole/properties/model/*

e.g., Linformer (path: /recbole/model/efficient_transformer_recommender/linformer.py),
Performer (path: /recbole/model/efficient_transformer_recommender/performer.py),
Synthesizer (path: /recbole/model/efficient_transformer_recommender/synthesizer.py), etc.