bert4rec

Pytorch implementation of BERT4Rec and Netflix VAE.

Sequence-to-Sequence Generative Model for Sequential Recommender System

common used Recommend Baseline Model, including the traditional statistical model, and the Nerual Network Model. Focus on the SRS (Sequential Recommend System).

Implementation of recommender ( Pytorch & Keras )

A comprehensive toolkit package designed to help you accurately predict key metrics in commercial area

PLUM-style Semantic-ID generative recommendation on MovieLens-1M with Qwen, LoRA/QLoRA, RQ-VAE, trie decoding, baselines, and reproducible evaluation

Two-stage sequential recommender on MovieLens 1M: BERT4Rec retrieval + DIN reranking + RRF fusion. Full-corpus HR@10 = 0.2974, NDCG@10 = 0.1692. Built from scratch in PyTorch.

This is a demonstration of the BERT4Rec model as a movies recommender

cross attention based BERT4Rec focusing on hyper-parameter settings, Usage, Cross-Attention Based BERT4Rec.

"Sequential recommender models" 3rd year coursework at Higher School of Economics

WSDM - KKBox's Music Recommendation Challenge

Choosing what content to send each prescriber next, by which channel, with messages that quote only approved claims

Maintained fork of a team streaming app: BERT4Rec/SASRec recommendations via ONNX Runtime, FastAPI with JWT auth, semantic search and Watch Together chat

This project is a reimplementation of the BERT4Rec model in PyTorch. The original BERT4Rec was implemented in TensorFlow.

Pet projects involving recommender systems

News click-through prediction over BERT article embeddings - NCF, NCF-Layer, and a Transformer (BERT4Rec) that scores the whole candidate slate jointly rather than item by item.