A single model unifies retrieval and context compression for on-device RAG via shared representations, matching traditional RAG performance at 1/10 context size with no extra storage.
Yongshuo Zong, Ondrej Bohdal, Tingyang Yu, Yongxin Yang, and Timothy Hospedales
9 Pith papers cite this work, alongside 46 external citations. Polarity classification is still indexing.
9
Pith papers citing it
46
external citations · external index
citation-role summary
background 2
citation-polarity summary
years
2026 9roles
background 2polarities
background 2representative citing papers
Unconstrained LLM rewriting of RDF dataset metadata maximizes retrieval gains but is least faithful; profile-grounded rewriting best balances effectiveness and grounding.
Agentic search over NASA EO-KG yields a 47k-pair benchmark where neural scoring plus LLM reranking raises MRR by over 5x then an additional 28%.
ReasonRec introduces a three-stage reasoning pipeline with CoT tuning, evidence-horizon curriculum, and uncertainty-guided delegation that claims over 30% gains in ranking metrics and up to 35% latency reduction on five recommendation datasets.
ClusterRAG applies density-based clustering to user profiles for collaborative retrieval in personalized RAG and reports best performance on LaMP tasks by combining target and similar-user profiles.
Synthetically formalizing information needs into topics with descriptions and narratives improves LLM relevance assessor agreement with humans and reduces over-labeling of relevant documents on TREC Deep Learning and Robust04.
Augmenting LLMs with bug references, few-shot learning, chain-of-thought, and RAG improves MPI error detection accuracy from 44% to 77% and generalizes across models.
TASTE dataset and MuQ-token aggregation enable effective use of audio features from large music models to improve content-based music recommendations over collaborative filtering alone.
RecNextEval is a reference implementation that applies time-window splits for temporal next-batch recommendation evaluation to minimize data leakage.
citing papers explorer
-
A Unified Model and Document Representation for On-Device Retrieval-Augmented Generation
A single model unifies retrieval and context compression for on-device RAG via shared representations, matching traditional RAG performance at 1/10 context size with no extra storage.
-
Faithful or Findable? Evaluating LLM-Generated Metadata for RDF Dataset Search
Unconstrained LLM rewriting of RDF dataset metadata maximizes retrieval gains but is least faithful; profile-grounded rewriting best balances effectiveness and grounding.
-
Bringing Agentic Search to Earth Observation Data Discovery
Agentic search over NASA EO-KG yields a 47k-pair benchmark where neural scoring plus LLM reranking raises MRR by over 5x then an additional 28%.
-
ReasonRec: A Reasoning-Augmented Multimodal Agent for Unified Recommendation
ReasonRec introduces a three-stage reasoning pipeline with CoT tuning, evidence-horizon curriculum, and uncertainty-guided delegation that claims over 30% gains in ranking metrics and up to 35% latency reduction on five recommendation datasets.
-
ClusterRAG: Cluster-Based Collaborative Filtering for Personalized Retrieval-Augmented Generation
ClusterRAG applies density-based clustering to user profiles for collaborative retrieval in personalized RAG and reports best performance on LaMP tasks by combining target and similar-user profiles.
-
Formalized Information Needs Improve Large-Language-Model Relevance Judgments
Synthetically formalizing information needs into topics with descriptions and narratives improves LLM relevance assessor agreement with humans and reduces over-labeling of relevant documents on TREC Deep Learning and Robust04.
-
Improving MPI Error Detection and Repair with Large Language Models and Bug References
Augmenting LLMs with bug references, few-shot learning, chain-of-thought, and RAG improves MPI error detection accuracy from 44% to 77% and generalizes across models.
-
Revisiting Content-Based Music Recommendation: Efficient Feature Aggregation from Large-Scale Music Models
TASTE dataset and MuQ-token aggregation enable effective use of audio features from large music models to improve content-based music recommendations over collaborative filtering alone.
-
RecNextEval: A Reference Implementation for Temporal Next-Batch Recommendation Evaluation
RecNextEval is a reference implementation that applies time-window splits for temporal next-batch recommendation evaluation to minimize data leakage.