AI Tools & Models HOTFeaturedAll
arXiv cs.AIScore60

RecMem: Recurrent Memory Compression for Long-Sequence Generative Recommendation

Summary

This work compresses long interaction histories with recurrent memory so generative recommenders can use long context without linearly growing cost.

Why this matters

Relevant to this topic

Open original ↗

Source: arXiv cs.AI · arxiv.org

Related items