# Training with Missed Targets in Generative Recommendation: Separating Supervision from Probability Competition

arXiv: https://arxiv.org/abs/2610.10124 · HTML: https://mlsanity.com/paper/2610.10124/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 12 of its 40 references that are in the map.
- First posted on arXiv 7 Oct 2026.

## Builds on (most cited first, 12 of 12)

- [OpenOneRec Technical Report](https://mlsanity.com/paper/2512.24762/) (2025)
- [Understanding Semantic IDs: From Item Representation to Item Selection in Generative Recommendation](https://mlsanity.com/paper/2607.24995/) (2026)
- [Can Generative Recommendation Reach Cold Items? A Temporal Perspective on Semantic-ID Generation](https://mlsanity.com/paper/2607.21101/) (2026)
- [UniRec: Bridging the Expressive Gap between Generative and Discriminative Recommendation via Chain-of-Attribute](https://mlsanity.com/paper/2604.12234/) (2026)
- [Fine-Tuned LLM as a Complementary Predictor Improving Ads System](https://mlsanity.com/paper/2605.27856/) (2026)
- [UniSGR: Unified Framework for Semantic ID Generation and Ranking](https://mlsanity.com/paper/2607.04068/) (2026)
- [Reward Guided Decoding for Generative Recommendation](https://mlsanity.com/paper/2607.25344/) (2026)
- [PSG: Pair-Space Generation for Efficient Generative Reranking](https://mlsanity.com/paper/2607.26427/) (2026)
- [Difficulty-Aware Semantic-ID Optimization for Generative Recommendation](https://mlsanity.com/paper/2608.20611/) (2026)
- [Off-Policy Evaluation for Semantic ID Recommenders: Does the Model's Own Code Hierarchy Help?](https://mlsanity.com/paper/2608.28905/) (2026)
- [History-Conditioned Joint-Prefix Alignment for Generative Recommendation](https://mlsanity.com/paper/2608.29179/) (2026)
- [Recommendation Retrievers Need Verifiers: Universal Generative Reranking for Sequential Recommendations](https://mlsanity.com/paper/2609.12270/) (2026)

## Abstract (arXiv)

Generative recommenders return a limited candidate set and may omit observed targets before reranking. A training strategy appends these missed targets to reranker training lists, although inference still ranks only original candidates. This operation simultaneously changes retrieved-target weight, adds supervision over appended targets, and makes the two groups compete for probability. An append/no-append comparison therefore cannot explain changes in returned-item rankings. We construct three matched losses that hold retrieved-target weight fixed while introducing appended-target supervision and group competition separately. The intermediate loss trains within both groups but normalizes them separately, preventing training-only targets from competing with inference candidates. Experiments with a released OneRec model and locally trained Amazon generators show that this competition can harm returned-item ranking. In four prespecified Amazon Video Games comparisons, removing it improved full-target normalized discounted cumulative gain (FT-NDCG) by 7.8--22.2%; 95% intervals over users and three of four intervals over training runs excluded zero. A conservative development-set rule selected appended-target training for two of three generators in one held-out category and rejected it for all three in another, avoiding a 1.7% loss. Candidate completion should therefore be evaluated for each generator rather than applied automatically.