A self-supervised, cluster-based hard negative sampling technique that draws negatives from the same semantic cluster as the positive item; in the production deployment the clusters are derived from large language model based multimodal content representations, so that intra-cluster items are genuinely similar and yiel...
Ivan Ji, Liu-Yi Hu, Harrison Zhao et al.· arXiv.org· 0 citations
Instead of rebuilding the entire recommendation stack from scratch, LIGE-GR generalizes the existing pointwise recommendation system into a listwise generation system, which allows mature recommender systems to benefit from listwise optimization while preserving compatibility with existing models, value functions, and...
V. Srinivas, Chen-Zhang He, Sam Woodmansee et al.· arXiv.org· 0 citations
A lightweight generative verifier for retrieval models is introduced, trained post hoc with next-token cross entropy, requires no sampled negatives or candidate pool during training, and scores only the retriever's top-$K$ candidates at inference.
Ben-Yu Zhang, Qiang Zhang, Rui Li et al.· 0 citations
Auto-research agents have shown the potential to automate hypothesis generation, experiment execution, and iterative refinement. However, scaling this paradigm to industry-scale recommendation models introduces two challenges: (1) long feedback loops, where model training can take days, making serial iteration prohibit...
Ming Li, Dai-Peng Li, Xu-Ying Ning et al.· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.