Closing the Indexing-Decoding Gap in Multimodal Generative Retrieval via Prefix Retention Optimization

• URL: http://arxiv.org/abs/2606.09241v2
• Date: Tue, 09 Jun 2026 08:19:13 GMT
• Title: Closing the Indexing-Decoding Gap in Multimodal Generative Retrieval via Prefix Retention Optimization
• Abstract summary: Multimodal generative retrieval formulates multimodal retrieval as discrete identifier generation, eliminating the need for explicit similarity search over external embeddings.<n>Existing approaches construct identifiers via residual quantization and decode them with trie-constrained beam search.<n>This combination introduces an indexing-decoding gap: identifier learning objectives, including reconstruction and contrastive losses, do not explicitly enforce prefix discriminability during decoding.<n>We propose PRO, a unified framework comprising three mechanisms: (i) prefix ranking distillation aligns quantized prefixes with those induced by pre-quantization embeddings using a listwise loss; (ii) vocabulary scheduling increases codebook
• Score: 68.48718919047127
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Multimodal generative retrieval formulates multimodal retrieval as discrete identifier generation, eliminating the need for explicit similarity search over external embeddings. Existing approaches construct identifiers via residual quantization and decode them with trie-constrained beam search. This combination introduces an indexing-decoding gap: identifier learning objectives, including reconstruction and contrastive losses, do not explicitly enforce prefix discriminability during decoding. As a result, even well-optimized identifiers can be irreversibly pruned early in beam search due to low-rank prefixes. We theoretically characterize this gap and derive a survival bound that relates prefix retention to three controllable factors in indexing and decoding. Building on this bound, we propose PRO, prefix retention optimization, a unified framework comprising three mechanisms: (i) prefix ranking distillation aligns quantized prefix rankings with those induced by pre-quantization embeddings using a listwise loss; (ii) vocabulary scheduling increases codebook sizes from shallow to deep residual quantization levels to reduce early competition from non-target prefixes; and (iii) geometric score fusion vectorizes each candidate prefix and incorporates its similarity to the query into beam search scoring, further reducing the indexing-decoding mismatch. Experiments on nine multimodal retrieval tasks show that PRO improves retention of target identifier prefixes and outperforms existing multimodal generative retrieval baselines.

Related papers

• WIDE: Wildcard Inference with Dynamic Expansion for Cross-Modal Generative Retrieval [11.681276883454288]
  We propose Wildcard Inference with Dynamic Expansion (WIDE) to overcome the problem of forced hallucination in cross-modal retrieval.<n>During the decoding generation phase, Asymmetry-aware Wildcard Decoding (AWD) detects semantic blind spots and emits wildcards instead of forced deterministic identifiers.<n>Blind-Spot Re-ranking (BSR) evaluates the expanded candidate pool using a hybrid scoring mechanism that combines discrete generation confidence with continuous semantic similarity.
  arXiv  Detail & Related papers  (2026-09-03T08:55:05Z)
• CodeHID: Learning an Addressable Hierarchical Code Index for Generative Code Retrieval [16.418964135916433]
  We propose CodeHID, a generative code retrieval framework that reformulates the code retrieval task from flat candidate matching to coarse-to-fine semantic address generation.<n>CodeHID relies on two core components. First, Pseudo-Neighbor Guided DocID Learning constructs a globally static hierarchical index by applying multi-level residual quantization and $k$-nearest-neighbor pseudo-supervision.<n>Second, Dual-Phase DocID Generation Guidance reliably navigates this fixed index by combining training-side ranking enhancements, using hard negatives and rank distillation, with inference-side candidate constraints and prefix-aware decoding.
  arXiv  Detail & Related papers  (2026-08-25T05:35:17Z)
• Beyond Fixed Depths and Widths: Optimizing Textual Decoding Tries in LLM-based Generative Recommendation [59.20704286413859]
  We propose a novel framework that co-designs textual term IDs and their underlying decoding trie.<n> BONSAI achieves up to a 21.6% relative improvement over state-of-the-art baselines.
  arXiv  Detail & Related papers  (2026-07-18T04:31:02Z)
• CAMI: Cost-Aware Agent-Guided Multi-Indexing for Semantic Retrieval [7.07006462273528]
  CAMI (Cost-Aware Multi-Indexing) is a framework that formalizes multi-index construction as a budgeted, multi-objective portfolio selection problem.<n>Our findings reveal that the framework systematically isolates high-recall portfolios under strict budget constraints.<n> CAMI is able to systematically identify these high-recall portfolios using up to 5x less budget compared to random search baselines.
  arXiv  Detail & Related papers  (2026-06-14T16:59:18Z)
• Decoupled Residual Quantization for Robust Semantic IDs in Recommendation [6.85237130729423]
  Poor quality tokenizer may come from codebook underutilization, unstable decision boundaries, or geometric distortion of the embedding space.<n>This paper develops a quantitative framework for diagnosing these failures through expected codeword overlap and effective codebook capacity.<n>As a proof of concept, we present Decoupled Residual Quantization (DRQ), which separates continuous geometry reconstruction from discrete distribution matching.
  arXiv  Detail & Related papers  (2026-06-01T07:55:21Z)
• Efficient Test-Time Inference via Deterministic Exploration of Truncated Decoding Trees [68.04613115686509]
  Self-consistency boosts inference-time performance by sampling multiple reasoning traces in parallel and voting.<n>We propose Distinct Leafion (DLE), a deterministic decoding method that treats truncated sampling as a pruned tree and systematically enumerates distinct leaves instead of sampling with replacement.<n>DLE explores higher-quality reasoning traces, yielding better performance on math, coding, and general reasoning tasks.
  arXiv  Detail & Related papers  (2026-04-22T12:42:03Z)
• Randomized Antipodal Search Done Right for Data Pareto Improvement of LLM Unlearning [32.47205256755341]
  Large language models (LLMs) sometimes memorize undesirable knowledge, which must be removed after deployment.<n>We propose Randomized Antipodal Search on Linearized Influence Kernel (RASLIK), a retrieval algorithm that combines permutation-projection hashing with randomized antipodal search.<n>Across multiple models, datasets, and unlearning algorithms, RASLIK consistently outperforms deterministic baselines and even oracle sampling.
  arXiv  Detail & Related papers  (2026-04-17T17:43:39Z)
• LLaDA-Rec: Discrete Diffusion for Parallel Semantic ID Generation in Generative Recommendation [32.284624021041004]
  We propose LLaDA-Rec, a discrete diffusion framework that reformulates recommendation as parallel semantic ID generation.<n> Experiments on three real-world datasets show that LLaDA-Rec consistently outperforms both ID-based and state-of-the-art generative recommenders.
  arXiv  Detail & Related papers  (2025-11-09T07:12:15Z)
• Constrained Auto-Regressive Decoding Constrains Generative Retrieval [71.71161220261655]
  Generative retrieval seeks to replace traditional search index data structures with a single large-scale neural network.<n>In this paper, we examine the inherent limitations of constrained auto-regressive generation from two essential perspectives: constraints and beam search.
  arXiv  Detail & Related papers  (2025-04-14T06:54:49Z)
• Generative Retrieval as Multi-Vector Dense Retrieval [71.75503049199897]
  Generative retrieval generates identifiers of relevant documents in an end-to-end manner. Prior work has demonstrated that generative retrieval with atomic identifiers is equivalent to single-vector dense retrieval. We show that generative retrieval and multi-vector dense retrieval share the same framework for measuring the relevance to a query of a document.
  arXiv  Detail & Related papers  (2024-03-31T13:29:43Z)
• Learning to Rank in Generative Retrieval [62.91492903161522]
  Generative retrieval aims to generate identifier strings of relevant passages as the retrieval target. We propose a learning-to-rank framework for generative retrieval, dubbed LTRGR. This framework only requires an additional learning-to-rank training phase to enhance current generative retrieval systems.
  arXiv  Detail & Related papers  (2023-06-27T05:48:14Z)
• Rank-Consistency Deep Hashing for Scalable Multi-Label Image Search [90.30623718137244]
  We propose a novel deep hashing method for scalable multi-label image search. A new rank-consistency objective is applied to align the similarity orders from two spaces. A powerful loss function is designed to penalize the samples whose semantic similarity and hamming distance are mismatched.
  arXiv  Detail & Related papers  (2021-02-02T13:46:58Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.