mlsanityData to 7 Oct 2026

xRAG: Extreme Context Compression for Retrieval-augmented Generation with One Token

arXiv 2405.13792 ↗First posted 22 May 2024cs.CL cs.AI cs.IRcited by 123

On the map: Retrieval-augmented generation → Generation and context integration

also in 1 more
Abstract (arXiv)

This paper introduces xRAG, an innovative context compression method tailored for retrieval-augmented generation. xRAG reinterprets document embeddings in dense retrieval--traditionally used solely for retrieval--as features from the retrieval modality. By employing a modality fusion methodology, xRAG seamlessly integrates these embeddings into the language model representation space, effectively eliminating the need for their textual counterparts and achieving an extreme compression rate. In xRAG, the only trainable component is the modality bridge, while both the retriever and the language model remain frozen. This design choice allows for the reuse of offline-constructed document embeddings and preserves the plug-and-play nature of retrieval augmentation. Experimental results demonstrate that xRAG achieves an average improvement of over 10% across six knowledge-intensive tasks, adaptable to various language model backbones, ranging from a dense 7B model to an 8x7B Mixture of Experts configuration. xRAG not only significantly outperforms previous context compression methods but also matches the performance of uncompressed models on several datasets, while reducing overall FLOPs by a factor of 3.53. Our work pioneers new directions in retrieval-augmented generation from the perspective of multimodality fusion, and we hope it lays the foundation for future efficient and scalable retrieval-augmented systems

Key facts. Cited by 123 papers in the map (data through 7 Oct 2026). Placed in Generation and context integration, one of 15 directions in Retrieval-augmented generation. Builds on 45 of its 70 references that are in the map. First posted on arXiv 22 May 2024.

Builds on

45 of 70 references.

Closest ancestors · 4

Same direction, newest first

  1. 2023RECOMP: Improving Retrieval-Augmented LMs with Compression and Selective Augmentation6 Oct
  2. 2022Demonstrate-Search-Predict: Composing retrieval and language models for knowledge-intensive NLP28 Dec
  3. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks22 May
  4. 2020REALM: Retrieval-Augmented Language Model Pre-Training10 Feb

Other references · 41

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2023Visual Instruction Tuningcited by 7,792
  3. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,069
  4. 2020Dense Passage Retrieval for Open-Domain Question Answeringcited by 3,779
  5. 2018HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answeringcited by 3,686
  6. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,895
  7. 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,793
  8. 2016MS MARCO: A Human Generated MAchine Reading COmprehension Datasetcited by 1,591
  9. 2018Know What You Don't Know: Unanswerable Questions for SQuADcited by 1,538
  10. 2017Get To The Point: Summarization with Pointer-Generator Networkscited by 1,281
  11. 2019PubMedQA: A Dataset for Biomedical Research Question Answeringcited by 1,271
  12. 2022Text Embeddings by Weakly-Supervised Contrastive Pre-trainingcited by 1,216
29 more
  1. 2020Leveraging Passage Retrieval with Generative Models for Open Domain Question Answeringcited by 1,203
  2. 2019DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphscited by 883
  3. 2019Generalization through Memorization: Nearest Neighbor Language Modelscited by 788
  4. 2017Reading Wikipedia to Answer Open-Domain Questionscited by 777
  5. 2018CoQA: A Conversational Question Answering Challengecited by 674
  6. 2023An Empirical Study of Catastrophic Forgetting in Large Language Models During Continual Fine-tuningcited by 622
  7. 2017The NarrativeQA Reading Comprehension Challengecited by 590
  8. 2020KILT: a Benchmark for Knowledge Intensive Language Taskscited by 571
  9. 2019SAMSum Corpus: A Human-annotated Dialogue Dataset for Abstractive Summarizationcited by 530
  10. 2023LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compressioncited by 443
  11. 2023How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resourcescited by 405
  12. 2023Learning to Compress Prompts with Gist Tokenscited by 379
  13. 2024LLMLingua-2: Data Distillation for Efficient and Faithful Task-Agnostic Prompt Compressioncited by 343
  14. 2024Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Modelscited by 336
  15. 2022Generate rather than Retrieve: Large Language Models are Strong Context Generatorscited by 296
  16. 2020Towards Understanding Ensemble, Knowledge Distillation and Self-Distillation in Deep Learningcited by 292
  17. 2023In-context Autoencoder for Context Compression in a Large Language Modelcited by 190
  18. 2023Text Embeddings Reveal (Almost) As Much As Textcited by 189
  19. 2023Compressing Context to Enhance Inference Efficiency of Large Language Modelscited by 185
  20. 2018Self-Imitation Learningcited by 135
  21. 2023How to Train Your DRAGON: Diverse Augmentation Towards Generalizable Dense Retrievalcited by 116
  22. 2022Training Language Models with Memory Augmentationcited by 112
  23. 2023Lift Yourself Up: Retrieval-augmented Text Generation with Self Memorycited by 103
  24. 2024Synthetic Data (Almost) from Scratch: Generalized Instruction Tuning for Language Modelscited by 99
  25. 2023Sentence Embedding Leaks More Information than You Expect: Generative Embedding Inversion Attack to Recover the Whole Sentencecited by 81
  26. 2022Training Data is More Valuable than You Think: A Simple and Effective Method by Retrieving from Training Datacited by 61
  27. 2023FactKG: Fact Verification via Reasoning on Knowledge Graphscited by 57
  28. 2024A Survey of Imitation Learning Methods, Environments and Metricscited by 16
  29. 2023Decouple knowledge from parameters for plug-and-play language modelingcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 16; 2025: 45; 2026: 62.

123 papers.

Sort:

By area:

  1. 2025Large Language Model Agent: A Survey on Methodology, Applications and Challengescited by 194
  2. 2024mPLUG-DocOwl2: High-resolution Compressing for OCR-free Multi-page Document Understandingcited by 113
  3. 2024InstructRAG: Instructing Retrieval-Augmented Generation via Self-Synthesized Rationalescited by 100
  4. 2024Prompt Compression for Large Language Models: A Surveycited by 65
  5. 2024CompAct: Compressing Retrieved Documents Actively for Question Answeringcited by 64
  6. 2024500xCompressor: Generalized Prompt Compression for Large Language Modelscited by 45
  7. 2026SWE-Pruner: Self-Adaptive Context Pruning for Coding Agentscited by 40
  8. 2025LongCodeZip: Compress Long Context for Code Language Modelscited by 38
  9. 2025A Survey of Multimodal Retrieval-Augmented Generationcited by 37
  10. 2025Rethinking Memory in LLM based Agents: Representations, Operations, and Emerging Topicscited by 36
  11. 2024Context Embeddings for Efficient Answer Generation in RAGcited by 26
  12. 2025Retrieval-Augmented Generation: A Comprehensive Survey of Architectures, Enhancements, and Robustness Frontierscited by 26
38 more
  1. 2025Dynamic Parametric Retrieval Augmented Generation for Test-time Knowledge Enhancementcited by 24
  2. 2024Perception Compressor: A Training-Free Prompt Compression Framework in Long Context Scenarioscited by 22
  3. 2025A Survey of Token Compression for Efficient Multimodal Large Language Modelscited by 22
  4. 2025DAST: Context-Aware Compression in LLMs via Dynamic Allocation of Soft Tokenscited by 21
  5. 2024Leveraging Passage Embeddings for Efficient Listwise Reranking with Large Language Modelscited by 20
  6. 2025PISCO: Pretty Simple Compression for Retrieval-Augmented Generationcited by 20
  7. 2025Retrieval Augmented Generation Evaluation in the Era of Large Language Models: A Comprehensive Surveycited by 20
  8. 2025GMSA: Enhancing Context Compression via Group Merging and Layer Semantic Alignmentcited by 20
  9. 2025AttentionRAG: Attention-Guided Context Pruning in Retrieval-Augmented Generationcited by 19
  10. 2025MLP Memory: A Retriever-Pretrained Memory for Large Language Modelscited by 19
  11. 2025DocPruner: A Storage-Efficient Framework for Multi-Vector Visual Document Retrieval via Adaptive Patch-Level Embedding Pruningcited by 19
  12. 2024FunnelRAG: A Coarse-to-Fine Progressive Retrieval Paradigm for RAGcited by 18
  13. 2024EXIT: Context-Aware Extractive Compression for Enhancing Retrieval-Augmented Generationcited by 17
  14. 2025Hierarchical Document Refinement for Long-context Retrieval-augmented Generationcited by 16
  15. 2024UncertaintyRAG: Span-Level Uncertainty Enhanced Long-Context Modeling for Retrieval-Augmented Generationcited by 14
  16. 2025Towards General Continuous Memory for Vision-Language Modelscited by 14
  17. 2025LastingBench: Defend Benchmarks Against Knowledge Leakagecited by 14
  18. 2025Understanding and Improving Information Preservation in Prompt Compression for LLMscited by 13
  19. 2026COMI: Coarse-to-fine Context Compression via Marginal Information Gaincited by 13
  20. 2026Read As Human: Compressing Context via Parallelizable Close Reading and Skimmingcited by 10
  21. 2025SARA: Selective and Adaptive Retrieval-augmented Generation with Context Compressioncited by 9
  22. 2025CLaRa: Bridging Retrieval and Generation with Continuous Latent Reasoningcited by 9
  23. 2025Text or Pixels? It Takes Half: On the Token Efficiency of Visual Text Inputs in Multimodal LLMscited by 8
  24. 2026Prompt Compression in the Wild: Measuring Latency, Rate Adherence, and Quality for Faster LLM Inferencecited by 8
  25. 2024From Reading to Compressing: Exploring the Multi-document Reader for Prompt Compressioncited by 7
  26. 2025Beyond Hard and Soft: Hybrid Context Compression for Balancing Local and Global Information Retentioncited by 7
  27. 2025Enhancing RAG Efficiency with Adaptive Context Compressioncited by 7
  28. 2025AdmTree: Compressing Lengthy Context with Adaptive Semantic Treescited by 7
  29. 2024CMT: A Memory Compression Method for Continual Knowledge Learning of Large Language Modelscited by 6
  30. 2025No Mean Feat: Simple, Strong Baselines for Context Compressioncited by 6
  31. 2026Gated Differentiable Working Memory for Long-Context Language Modelingcited by 6
  32. 2026Cartridges at Scale: Training Modular KV Caches over Large Document Collectionscited by 6
  33. 2025Conflict-Aware Soft Prompting for Retrieval-Augmented Generationcited by 5
  34. 2025CompLLM: Compression for Long Context Q&Acited by 5
  35. 2026Can Vision-Language Models Handle Long-Context Code? An Empirical Study on Visual Compressioncited by 5
  36. 2026ATACompressor: Adaptive Task-Aware Compression for Efficient Long-Context Processing in LLMscited by 5
  37. 2025Sign Language Translation using Frame and Event Stream: Benchmark Dataset and Algorithmscited by 4
  38. 2025OSCAR: Online Soft Compression And Rerankingcited by 4

50 most cited of 123 All 123 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024Contextual Compression in Retrieval-Augmented Generation for Large Language Models: A Survey
  2. 2024Lighter And Better: Towards Flexible Context Adaptation For Retrieval Augmented Generation
  3. 2024XRAG: eXamining the Core -- Benchmarking Foundational Components in Advanced Retrieval-Augmented Generation
  4. 2025UniversalRAG: Retrieval-Augmented Generation over Corpora of Diverse Modalities and Granularities
  5. 2025MacRAG: Compress, Slice, and Scale-up for Multi-Scale Adaptive Context RAG
  6. 2026Rethinking Soft Compression in Retrieval-Augmented Generation: A Query-Conditioned Selector Perspective
  7. 2025ELITE: Embedding-Less retrieval with Iterative Text Exploration
  8. 2025Less Is More: Elevating RAG via Performance-Driven Context Compression