Year: 2025
Type: preprint
Abstract: Is bigger always better for time series foundation models? With the question in mind, we explore an alternative to training a single, large monolithic model: building a portfolio of smaller, pretrained forecasting models. By applying ensembling or model selection over these portfolios, we achieve competitive performance on large-scale benchmarks using much fewer parameters. We explore strategies for designing such portfolios and find that collections of specialist models consistently outperform portfolios of independently trained generalists. Remarkably, we demonstrate that post-training a base model is a compute-effective approach for creating sufficiently diverse specialists, and provide evidences that ensembling and model selection are more compute-efficient than test-time fine-tuning.
Source: arXiv (Cornell University)
Authors: Kayaalp, Mert, Turkmen, Caner, Shchur, Oleksandr, Mercado, Pedro, Ansari, Abdul Fatir, Bohlke-Schneider, Michael, Wang, Bernie
Language: English
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4415315032
Cites:
Cited by:
Related to: