Research map / Time series / Foundation models for time series
ExoTimer: Leveraging Large Language Models for Time Series Forecasting with Exogenous Variables
Lan Wu, Xuebin Wang, Chenglong Ge, Ruijuan Chu, LinYu Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add LSH-based Bi-Hash Alignment between text and temporal modalities (ablation) | ExoTimer · ECL | MSE: 0.994 -> 0.317 | Improved |
| add exogenous variable information to endogenous embedding (ablation) | ExoTimer · ECL | MSE: 0.392 -> 0.317 | Improved |
| select important exogenous variables adaptively with mask (ablation) | ExoTimer · PJM | MSE: 0.127 -> 0.063 | Improved |
| add adaptive weighting in dual-branch predictor (ablation) | ExoTimer · PJM | MSE: 0.114 -> 0.063 | Improved |
| add both text-temporal and temporal-text alignment branches vs either alone (ablation) | ExoTimer · ECL | MSE: 0.394 / 0.379 -> 0.317 | Improved |
| use multi-attribute prompts instead of ablated prompts or raw numeric text prompts (ablation) | ExoTimer · ETTh1, Weather | MSE | Improved |
| ExoTimer vs ExoLLM on long-term forecasting | ExoTimer · Traffic | MSE: 0.145 -> 0.150 | Hurt |
| ExoTimer vs ExoLLM in few-shot (10% data) | ExoTimer · ETTh1 | MSE: 0.084 -> 0.076 | Improved |
| ExoTimer vs ExoLLM in zero-shot transfer | ExoTimer · ETTm1→ETTm2 | MSE: 0.162 -> 0.126 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TimeXer — TimeXer: Empowering Transformers for Time Series Forecasting with Exogenous Variables (NeurIPS 2024) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TimeCMA: Towards LLM-Empowered Multivariate Time Series Forecasting via Cross-Modality Alignment (AAAI 2025 · similarity 0.91)
- $S^2$IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting (ICML 2024 · similarity 0.90)
- TimeXer: Empowering Transformers for Time Series Forecasting with Exogenous Variables (NeurIPS 2024 · similarity 0.90)
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024 · similarity 0.90)
- TimeXL: Explainable Multi-modal Time Series Prediction with LLM-in-the-Loop (NeurIPS 2025 · similarity 0.90)
- AutoTimes: Autoregressive Time Series Forecasters via Large Language Models (NeurIPS 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
