REVIEW 11 cited by
Ultron: An Ultimate Retriever on Corpus with a Model-based Indexer
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Document retrieval has been extensively studied within the index-retrieve framework for decades, which has withstood the test of time. Unfortunately, such a pipelined framework limits the optimization of the final retrieval quality, because indexing and retrieving are separated stages that can not be jointly optimized in an end-to-end manner. In order to unify these two stages, we explore a model-based indexer for document retrieval. Concretely, we propose Ultron, which encodes the knowledge of all documents into the model and aims to directly retrieve relevant documents end-to-end. For the model-based indexer, how to represent docids and how to train the model are two main issues to be explored. Existing solutions suffer from semantically deficient docids and limited supervised data. To tackle these two problems, first, we devise two types of docids that are richer in semantics and easier for model inference. In addition, we propose a three-stage training workflow to capture more knowledge contained in the corpus and associations between queries and docids. Experiments on two public datasets demonstrate the superiority of Ultron over advanced baselines for document retrieval.
Forward citations
Cited by 11 Pith papers
-
Beyond Matching: Category-Guided Latent Intent Reasoning for Generative Retrieval in E-Commerce
CaLIR learns continuous latent intent states guided by product category hierarchies for generative retrieval, combining hierarchical reasoning and dynamic prefix tries to balance effectiveness and low-latency inferenc...
-
A Parametric Memory Head for Continual Generative Retrieval
A product-key parametric memory head with selective sparse updates mitigates catastrophic forgetting in generative retrieval models during sequential addition of new documents.
-
Model Editing for New Document Integration in Generative Information Retrieval
DOME adapts generative IR models to unseen documents via critical-layer identification, hybrid-label edit vector optimization, and parameter updates, achieving strong new-document retrieval with reduced training cost.
-
Generative Chinese Statute Retrieval
A generative retriever with multi-granularity structured statute IDs and multi-task training outperforms strong sparse, dense, and legal baselines on the STARD Chinese statute benchmark.
-
Integrating Chain-of-Thought into Generative Retrieval: A Preliminary Study
ThinkGR interleaves chain-of-thought with docid generation using hybrid decoding and two-phase training to achieve state-of-the-art results on multi-hop retrieval benchmarks.
-
Lost in Decoding? Reproducing and Stress-Testing the Look-Ahead Prior in Generative Retrieval
Reproduction confirms PAG boosts generative retrieval effectiveness, but its look-ahead planning signal collapses under intent-preserving typos and query mismatches, reverting performance to unguided decoding.
-
Towards Efficient and Generalizable Retrieval: Adaptive Semantic Quantization and Residual Knowledge Transfer
SA²CRQ uses sequential adaptive residual quantization based on path entropy plus anchored curriculum regularization from head items to improve both efficiency and cold-start performance in generative retrieval.
-
MixLoRA-DSI: Dynamically Expandable Mixture-of-LoRA Experts for Rehearsal-Free Generative Retrieval over Dynamic Corpora
A rehearsal-free generative retriever that expands mixture-of-LoRA experts only when router energy scores flag out-of-distribution tokens, achieving sublinear parameter growth over dynamic corpora.
-
Why These Documents? Explainable Generative Retrieval with Hierarchical Category Paths
HyPE improves generative retrieval by first generating hierarchical category paths for explainability and then using path-aware ranking to boost performance.
-
UniGD: A Unified Generative-Discriminative Framework for Industrial Retrieval
UniGD couples generative retrieval with explicit relevance scoring in one model, reporting +5.78% ad revenue, 33.1% lower latency at Kuaishou, and improved Recall@10 on NQ320K and MS300K.
-
Understanding and Debugging Failures in N-Gram-Based Generative Retrieval
Presents a taxonomy of generative retrieval failures, empirically identifies issues such as ambiguous docids and low diversity in n-gram methods, and introduces a web-based debugging tool.
Discussion (0). Sign in to comment.