1 paper
Tianhe Lin, Ziwei Xiong, Baoyuan Ou +8
Modeling ultra-long user behavior sequences is pivotal for capturing evolving and lifelong interests in modern recommendation systems. However, deploying such models in real-time i…