# Guiding the coarse levels of semantic IDs makes the fine levels learnable

arXiv: https://arxiv.org/abs/2609.22227 · HTML: https://mlsanity.com/paper/2609.22227/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 16 of its 20 references that are in the map.
- First posted on arXiv 3 Sep 2026.

## Builds on (most cited first, 16 of 16)

- [Adapting Large Language Models by Integrating Collaborative Semantics for Recommendation](https://mlsanity.com/paper/2311.09049/) (2023)
- [Learnable Item Tokenization for Generative Recommendation](https://mlsanity.com/paper/2405.07314/) (2024)
- [PLUM: Adapting Pre-trained Language Models for Industrial-scale Generative Recommendations](https://mlsanity.com/paper/2510.07784/) (2025)
- [Differentiable Semantic ID for Generative Recommendation](https://mlsanity.com/paper/2601.19711/) (2026)
- [REG4Rec: Reasoning-Enhanced Generative Model for Large-Scale Recommendation Systems](https://mlsanity.com/paper/2508.15308/) (2025)
- [Reasoning over Semantic IDs Enhances Generative Recommendation](https://mlsanity.com/paper/2603.23183/) (2026)
- [HiD-VAE: Interpretable Generative Recommendation via Hierarchical and Disentangled Semantic IDs](https://mlsanity.com/paper/2508.04618/) (2025)
- [GPR: Towards a Generative Pre-trained One-Model Paradigm for Large-Scale Advertising Recommendation](https://mlsanity.com/paper/2511.10138/) (2025)
- [PinRec: Unified Generative Retrieval for Pinterest Recommender Systems](https://mlsanity.com/paper/2504.10507/) (2025)
- [Generative Recommendation for Large-Scale Advertising](https://mlsanity.com/paper/2602.22732/) (2026)
- [Stop Treating Collisions Equally: Qualification-Aware Semantic ID Learning for Recommendation at Industrial Scale](https://mlsanity.com/paper/2603.00632/) (2026)
- [OneRanker: Unified Generation and Ranking with One Model in Industrial Advertising Recommendation](https://mlsanity.com/paper/2603.02999/) (2026)
- [Bi-Level Optimization for Generative Recommendation: Bridging Tokenization and Generation](https://mlsanity.com/paper/2510.21242/) (2025)
- [Taming the Long Tail: Denoising Collaborative Information for Robust Semantic ID Generation](https://mlsanity.com/paper/2510.25622/) (2025)
- [LLMs Need Encoders for Semantic IDs Too](https://mlsanity.com/paper/2606.00324/) (2026)
- [FedMM: Federated Collaborative Signal Quantization for Multi-Market CTR Prediction](https://mlsanity.com/paper/2605.11433/) (2026)

## Abstract (arXiv)

Generative retrieval represents each item by a short Semantic ID and casts recommendation as autoregressive generation of that sequence. Because the tokenizer is trained independently to reconstruct an item embedding, its codes are aligned with neither the downstream LLM nor the end task. Nearly every SID system therefore spends extra effort to bridge this gap--alignment corpora, reasoning/RL, or per-token encoders to make codes legible, or learned tokenizer supervision to make them task-aware--yet the recovered meaning is content-derived and may not be the meaning the task needs. We introduce Guided SID, which instead makes the levels that matter most meaningful by construction: we force the coarse RQ-VAE levels to encode a predefined categorical attribute--chosen to be text-grounded (hence legible to the LLM) and task-relevant--by deterministic supervised index assignment (overriding nearest-neighbor selection with the attribute label) while keeping the codebooks learnable (they still receive reconstruction gradients). A trie-merge construction maps any high-cardinality or set-valued attribute onto the fixed code budget while keeping merged buckets semantically coherent. Guiding costs nothing intrinsically: collision and reconstruction match or beat the vanilla baseline despite pinning the coarse level. In a matched end-to-end A/B differing only in the SID encoding, the guided retriever improves recall@k at every list length we measure (1.36x at k=1, 1.39x at k=10), raises mean reciprocal rank from 0.0260 to 0.0355, and predicts the pre-defined attribute 4.2x more often.