# Accelerating Time Series Foundation Models with Speculative Decoding

arXiv: https://arxiv.org/abs/2511.18191 · HTML: https://mlsanity.com/paper/2511.18191/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Speculative decoding, one of 12 directions in Efficient LLM inference and serving.
- Builds on 18 of its 26 references that are in the map.
- First posted on arXiv 22 Nov 2025.

## Placed in

- Efficient LLM inference and serving / [Speculative decoding](https://mlsanity.com/inference/speculative-decoding/)

## Builds on (most cited first, 18 of 18)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting](https://mlsanity.com/paper/2106.13008/) (2021)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks](https://mlsanity.com/paper/1704.04110/) (2017)
- [Accelerating Large Language Model Decoding with Speculative Sampling](https://mlsanity.com/paper/2302.01318/) (2023)
- [Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networks](https://mlsanity.com/paper/1703.07015/) (2017)
- [Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads](https://mlsanity.com/paper/2401.10774/) (2024)
- [N-BEATS: Neural basis expansion analysis for interpretable time series forecasting](https://mlsanity.com/paper/1905.10437/) (2019)
- [Unified Training of Universal Time Series Forecasting Transformers](https://mlsanity.com/paper/2402.02592/) (2024)
- [EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees](https://mlsanity.com/paper/2406.16858/) (2024)
- [SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction](https://mlsanity.com/paper/2106.09305/) (2021)
- [Blockwise Parallel Decoding for Deep Autoregressive Models](https://mlsanity.com/paper/1811.03115/) (2018)
- [Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts](https://mlsanity.com/paper/2409.16040/) (2024)
- [Sundial: A Family of Highly Capable Time Series Foundation Models](https://mlsanity.com/paper/2502.00816/) (2025)
- [TiRex: Zero-Shot Forecasting Across Long and Short Horizons with Enhanced In-Context Learning](https://mlsanity.com/paper/2505.23719/) (2025)
- [Continuous Speculative Decoding for Autoregressive Image Generation](https://mlsanity.com/paper/2411.11925/) (2024)
- [TPP-SD: Accelerating Transformer Point Process Sampling with Speculative Decoding](https://mlsanity.com/paper/2507.09252/) (2025)

## Built on by (most cited first, 1 of 1)

- [Accelerating Speculative Diffusions via Block Verification](https://mlsanity.com/paper/2606.13426/) (2026)

## Abstract (arXiv)

Time series forecasting drives operational decisions under tight latency budgets, and autoregressive time series foundation models (TSFMs) increasingly deliver the most accurate forecasts. That accuracy is paid for at inference, since a horizon of H steps takes lceil H / Prceil sequential forward passes of a large model, so latency grows with exactly the long horizons these models are prized for. Yet a far cheaper model predicts most next patches nearly as well as the large one, and causal models can verify a block of future patches in one parallel pass even though they generate them one at a time. These are precisely the conditions under which speculative decoding thrives in LLMs, but its ingredients are all defined over discrete vocabularies. We therefore develop speculative decoding for continuous patch autoregression. A cheap draft proposes K future patches, and the target verifies all of them in a single causal pass, accepting each by a log-domain Gaussian likelihood-ratio test and correcting the first rejection with its own prediction. We prove that the accelerated output stays within a squared-error radius of target-only decoding set by an acceptance temperature, and that throughput follows a capped-geometric law that makes speedups predictable before deployment. The method delivers up to 3.0 × inference speedup at accuracy between target and draft across five TSFM families, and we characterize which architectures admit single-pass verification and when speculation does not pay.