mlsanityData to 14 Oct 2026

TS-RAG: Retrieval-Augmented Generation based Time Series Foundation Models are Stronger Zero-Shot Forecaster

arXiv 2503.07649 ↗First posted 6 Mar 2025cs.LG cs.AIcited by 40

On the map: Retrieval-augmented generation → Time series and agricultural RAG

Abstract (arXiv)

Large Language Models (LLMs) and Foundation Models (FMs) have recently become prevalent for time series forecasting tasks. While fine-tuning LLMs enables domain adaptation, they often struggle to generalize across diverse and unseen datasets. Moreover, existing Time Series Foundation Models (TSFMs) still face challenges in handling non-stationary dynamics and distribution shifts, largely due to the lack of effective mechanisms for adaptation. To this end, we present TS-RAG, a retrieval-augmented generation framework for time series forecasting that enhances the generalization and interpretability of TSFMs. Specifically, TS-RAG leverages pre-trained time series encoders to retrieve semantically relevant segments from a dedicated knowledge base, enriching the contextual representation of the input query. Furthermore, we propose an Adaptive Retrieval Mixer (ARM) module that dynamically fuses the retrieved patterns with the TSFM's internal representation, improving forecasting accuracy without requiring task-specific fine-tuning. Thorough empirical studies on seven public benchmark datasets demonstrate that TS-RAG achieves state-of-the-art zero-shot forecasting performance, outperforming the existing TSFMs by up to 6.84% across diverse domains while also providing desirable interpretability. Our code and data are available at: https://github.com/UConn-DSIS/TS-RAG

Key facts. Cited by 40 papers in the map (data through 14 Oct 2026). Placed in Time series and agricultural RAG, one of 15 directions in Retrieval-augmented generation. Builds on 44 of its 57 references that are in the map. First posted on arXiv 6 Mar 2025.

Builds on

44 of 57 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024TimeRAG: BOOSTING LLM Time Series Forecasting via Retrieval-Augmented Generation21 Dec
  2. 2024Retrieval Augmented Time Series Forecasting12 Nov
  3. 2024Retrieval-Augmented Diffusion Models for Time Series Forecasting24 Oct

Other references · 41

  1. 2017Attention Is All You Needcited by 60,361
  2. 2020Language Models are Few-Shot Learnerscited by 32,419
  3. 2023GPT-4 Technical Reportcited by 18,396
  4. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  5. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,132
  6. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  7. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  8. 2016XGBoost: A Scalable Tree Boosting Systemcited by 5,166
  9. 2014Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translationcited by 4,057
  10. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,509
  11. 2020Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecastingcited by 2,169
  12. 2022TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysiscited by 1,058
29 more
  1. 2017Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networkscited by 879
  2. 2024Chronos: Learning the Language of Time Seriescited by 759
  3. 2023Time-LLM: Time Series Forecasting by Reprogramming Large Language Modelscited by 700
  4. 2023A decoder-only foundation model for time-series forecastingcited by 692
  5. 2023One Fits All:Power General Time Series Analysis by Pretrained LMcited by 632
  6. 2024Unified Training of Universal Time Series Forecasting Transformerscited by 614
  7. 2023Large Language Models Are Zero-Shot Time Series Forecasterscited by 500
  8. 2024MOMENT: A Family of Open Time-series Foundation Modelscited by 482
  9. 2024A Survey on RAG Meeting LLMs: Towards Retrieval-Augmented Large Language Modelscited by 452
  10. 2021Spectral Temporal Graph Neural Network for Multivariate Time-series Forecastingcited by 299
  11. 2024Foundation Models for Time Series Analysis: A Tutorial and Surveycited by 239
  12. 2024G-Retriever: Retrieval-Augmented Generation for Textual Graph Understanding and Question Answeringcited by 223
  13. 2024Timer: Generative Pre-trained Transformers Are Large Time Series Modelscited by 208
  14. 2016Temporal Convolutional Networks: A Unified Approach to Action Segmentationcited by 199
  15. 2025Sundial: A Family of Highly Capable Time Series Foundation Modelscited by 180
  16. 2023Fine-Tuning or Retrieval? Comparing Knowledge Injection in LLMscited by 178
  17. 2022PromptCast: A New Prompt-based Learning Paradigm for Time Series Forecastingcited by 172
  18. 2021Discrete Graph Structure Learning for Forecasting Multiple Time Seriescited by 171
  19. 2023TEST: Text Prototype Aligned Embedding to Activate LLM's Ability for Time Seriescited by 169
  20. 2024Are Language Models Actually Useful for Time Series Forecasting?cited by 168
  21. 2023TimeGPT-1cited by 158
  22. 2023LLM4TS: Aligning Pre-Trained LLMs as Data-Efficient Time-Series Forecasterscited by 129
  23. 2024Tiny Time Mixers (TTMs): Fast Pre-trained Models for Enhanced Zero/Few-Shot Forecasting of Multivariate Time Seriescited by 127
  24. 2024VisionTS: Visual Masked Autoencoders Are Free-Lunch Zero-Shot Time Series Forecasterscited by 88
  25. 2024Timer-XL: Long-Context Transformers for Unified Time Series Forecastingcited by 87
  26. 2024Empowering Time Series Analysis with Large Language Models: A Surveycited by 60
  27. 2022Retrieval Based Time Series Forecastingcited by 17
  28. 2024LLMs Know What They Need: Leveraging a Missing Information Guided Framework to Empower Retrieval-Augmented Generationcited by 17
  29. 2023Deep Learning for Energy Time-Series Analysis and Forecastingcited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 9; 2026: 31.

40 papers.

Sort:

By area:

  1. 2025Multi-modal Time Series Analysis: A Tutorial and Surveycited by 24
  2. 2026AutoSOTA: An End-to-End Automated Research System for State-of-the-Art AI Model Discoverycited by 23
  3. 2025Time-IMM: A Dataset and Benchmark for Irregular Multimodal Multivariate Time Seriescited by 19
  4. 2025MicroRemed: Benchmarking LLMs in Microservices Remediationcited by 11
  5. 2025TRACE: Grounding Time Series in Context for Multimodal Embedding and Retrievalcited by 7
  6. 2026Enhancing Multivariate Time Series Forecasting with Global Temporal Retrievalcited by 5
  7. 2026Stationarity-Aware Retrieval-Augmented Time Series Forecastingcited by 5
  8. 2026Time Series Reasoning via Process-Verifiable Thinking Data Synthesis and Scheduling for Tailored LLM Reasoningcited by 3
  9. 2026Channel-wise Retrieval for Multivariate Time Series Forecastingcited by 3
  10. 2026Post-Training in Time Series Foundation Models: A Unifying Frameworkcited by 3
  11. 2025LLM-based Question-Answer Framework for Sensor-driven HVAC System Interactioncited by 2
  12. 2025Anticipatory Evaluation of Language Models: Predicting LLM Benchmark Scores from Descriptionscited by 2
28 more
  1. 2025Towards Interpretable and Trustworthy Time Series Reasoning: A BlueSky Visioncited by 2
  2. 2026MEMTS: Internalizing Domain Knowledge via Parameterized Memory for Retrieval-Free Domain Adaptation of Time Series Foundation Modelscited by 2
  3. 2026Adapting Time Series Foundation Models through Data Mixturescited by 2
  4. 2026Semantics-Enhanced Retrieval-Augmented Time Series Forecastingcited by 2
  5. 2025Towards Temporal Knowledge-Base Creation for Fine-Grained Opinion Analysis with Language Modelscited by 1
  6. 2026Retrieval-Augmented Generation with Covariate Time Seriescited by 1
  7. 2026LLM-Oriented Information Retrieval: A Denoising-First Perspectivecited by 1
  8. 2026RareCP: Regime-Aware Retrieval for Efficient Conformal Predictioncited by 1
  9. 2026Factorize to Generalize: Retrieval-Guided Invariant-Dynamic Decomposition for Time Series Forecastingcited by 1
  10. 2026ReDiTT: Retrieval Augmented Conditional Diffusion Transformers for Asynchronous Time Seriescited by 1
  11. 2025STRATA-TS: Selective Knowledge Transfer for Urban Time Series Forecasting with Retrieval-Guided Reasoningcited by 0
  12. 2026Dual-Prototype Disentanglement: A Context-Aware Enhancement Framework for Time Series Forecastingcited by 0
  13. 2026Retrieval-Augmented Multi-scale Framework for County-Level Crop Yield Prediction Across Large Regionscited by 0
  14. 2026Beyond Extrapolation: Knowledge Utilization Paradigm with Bidirectional Inspiration for Time Series Forecastingcited by 0
  15. 2026CausalMoE: A Billion-Scale Multimodal Foundation Model for Granger Causal Discovery with Pattern-Routed Heterogeneous Expertscited by 0
  16. 2026Using LLMs for Explainable, Data-Driven Insight Generation from Time Seriescited by 0
  17. 2026Align-RAG: Alignment Is All You Need for TSFM In-Context Learningcited by 0
  18. 2026Model-agnostic Retrieval-Augmented Extended Forecasting for time seriescited by 0
  19. 2026Which Histories Matter for Time Series Forecasting? Learning Predictive Relevance with Future Supervisioncited by 0
  20. 2026When Does Retrieval Help Time-Series Forecasting?cited by 0
  21. 2026Tracing the Evidence Behind Zero-Shot Time-Series Forecasting: A Source-First Taxonomy and Audit Frameworkcited by 0
  22. 2026XMatch: Enhancing Covariate-Aware Time Series Forecasting through Tree-Structured Exogenous Matchingcited by 0
  23. 2026GNA: Granular Neighbor Assembly for Retrieval-Augmented Multivariate Time-Series Forecastingcited by 0
  24. 2026Raw-Routed Mixture of Adapters: A Causal Intervention for Routing Collapse in Time Series Foundation Modelscited by 0
  25. 2026Dual-Context Analog Retrieval for Time Series Forecastingcited by 0
  26. 2026GRAM: Correcting Frozen Time-Series Foundation Models via Graph-Retrieved Amplitude Memorycited by 0
  27. 2026AdaCast: Conditional Parameter Generation for Adaptive Time Series Forecastingcited by 0
  28. 2026ECHO: Correcting Time Series Foundation Model Forecasts with Historical Analogues and Errorscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026TS-RAG: Retrieval Augmented Generation for Time Series Forecasting
  2. 2026Not All Retrievals are Useful: Cross-Attention for Input-Aware RAG in Time Series Forecasting
  3. 2024TimeRAF: Retrieval-Augmented Foundation model for Zero-shot Time Series Forecasting
  4. 2025QuiZSF: A Retrieval-Augmented Framework for Zero-Shot Time Series Forecasting
  5. 2025Retrieval-augmented Large Language Models for Financial Time Series Forecasting
  6. 2026Beyond Similarity: Coverage-Aware Prompt Selection for Time Series Forecasting with LLMs
  7. 2025Retrieval Augmented Time Series Forecasting
  8. 2026RATL: Learning from Retrieved Residuals for Robust Multivariate Time-Series Forecasting