{"ID":6620444,"CreatedAt":"2026-07-15T01:01:48.440468303Z","UpdatedAt":"2026-07-15T03:28:55.185153975Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.12277","arxiv_id":"2607.12277","title":"Not Only NTP: Extending Training Signal Coverage for Generative Recommendation","abstract":"Next-Token Prediction (NTP) carries two structural training signal limitations. First, NTP optimizes for single-step prediction only, placing no supervised pressure on learning longer-range behavioral structure -- we term this \\textbf{temporal locality}. Second, in multi-domain sequences, each target item embedding receives gradient updates exclusively from the immediately preceding hidden state, with no explicit gradient pathway from cross-domain context -- we term this \\textbf{spatial locality}. We propose \\textbf{NONTP}, extending NTP's signal coverage along both dimensions through two auxiliary objectives. \\textbf{TCL (Temporal Contrastive Learning)} uses a BYOL-style EMA teacher with InfoNCE to align hidden states against a $K$-step future trajectory in representation space. \\textbf{TDL (Trans-Domain Learning)} mean-pools cross-domain hidden states and predicts through the shared prediction head, opening a second gradient pathway with no additional parameters. Both are discarded at inference: zero overhead. On a four-domain Meituan industrial dataset (full ranking), NONTP achieves HR@10 +34.3\\% over NTP and +18.3\\% over MBGR. On the public Amazon Movie-Book-CDs benchmark, HR@10 +2.8\\% and NDCG@10 +3.7\\%. Online A/B tests confirm CTR +1.8\\% and GMV +2.1\\% (both $p \u003c 0.01$). Ablation studies confirm each component contributes independently, with gradient conflict analyzed as a direction for future work.","short_abstract":"Next-Token Prediction (NTP) carries two structural training signal limitations. First, NTP optimizes for single-step prediction only, placing no supervised pressure on learning longer-range behavioral structure -- we term this \\textbf{temporal locality}. Second, in multi-domain sequences, each target item embedding rec...","url_abs":"https://arxiv.org/abs/2607.12277","url_pdf":"https://arxiv.org/pdf/2607.12277v1","authors":"[\"Changhao Li\",\"Shuli Wang\",\"Junwei Yin\",\"Senjie Kou\",\"Yinqiu Huang\",\"Chi Wang\",\"Yinhua Zhu\",\"Haitao Wang\",\"Xingxing Wang\"]","published":"2026-07-14T02:28:23Z","proceeding":"cs.IR","tasks":"[\"cs.IR\"]","methods":"[]","has_code":false}