Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FMTK: A Modular Toolkit for Composable Time Series Foundation Model Pipelines

Hetvi Shastri, Pragya Sharma, Walid A. Hanafy, Mani Srivastava, Prashant Shenoy

TL;DR

FMTK addresses the fragmentation in TSFM pipeline development by providing a standardized, modular toolkit that couples encoders, backbones, adapters, and decoders through a unified API. By decoupling components and enforcing consistent execution and evaluation semantics, it enables rapid, reproducible experimentation across diverse tasks and backbones with minimal code. The framework demonstrates usability, architectural adaptability, and competitive performance, including low overhead and memory benefits, while supporting runtime component switching and comprehensive benchmarking. This work offers a practical template for rigorously evaluating TSFM-based systems and lays groundwork for extending modular pipelines to broader foundation-model workflows.

Abstract

Foundation models (FMs) have opened new avenues for machine learning applications due to their ability to adapt to new and unseen tasks with minimal or no further training. Time-series foundation models (TSFMs) -- FMs trained on time-series data -- have shown strong performance on classification, regression, and imputation tasks. Recent pipelines combine TSFMs with task-specific encoders, decoders, and adapters to improve performance; however, assembling such pipelines typically requires ad hoc, model-specific implementations that hinder modularity and reproducibility. We introduce FMTK, an open-source, lightweight and extensible toolkit for constructing and fine-tuning TSFM pipelines via standardized backbone and component abstractions. FMTK enables flexible composition across models and tasks, achieving correctness and performance with an average of seven lines of code. https://github.com/umassos/FMTK

FMTK: A Modular Toolkit for Composable Time Series Foundation Model Pipelines

TL;DR

FMTK addresses the fragmentation in TSFM pipeline development by providing a standardized, modular toolkit that couples encoders, backbones, adapters, and decoders through a unified API. By decoupling components and enforcing consistent execution and evaluation semantics, it enables rapid, reproducible experimentation across diverse tasks and backbones with minimal code. The framework demonstrates usability, architectural adaptability, and competitive performance, including low overhead and memory benefits, while supporting runtime component switching and comprehensive benchmarking. This work offers a practical template for rigorously evaluating TSFM-based systems and lays groundwork for extending modular pipelines to broader foundation-model workflows.

Abstract

Foundation models (FMs) have opened new avenues for machine learning applications due to their ability to adapt to new and unseen tasks with minimal or no further training. Time-series foundation models (TSFMs) -- FMs trained on time-series data -- have shown strong performance on classification, regression, and imputation tasks. Recent pipelines combine TSFMs with task-specific encoders, decoders, and adapters to improve performance; however, assembling such pipelines typically requires ad hoc, model-specific implementations that hinder modularity and reproducibility. We introduce FMTK, an open-source, lightweight and extensible toolkit for constructing and fine-tuning TSFM pipelines via standardized backbone and component abstractions. FMTK enables flexible composition across models and tasks, achieving correctness and performance with an average of seven lines of code. https://github.com/umassos/FMTK

Paper Structure

This paper contains 13 sections, 1 figure, 2 tables.

Table of Contents

  1. Introduction
  2. Design
  3. Components
  4. Pipeline Abstraction
  5. Runtime Metrics and Benchmarking Support
  6. Evaluation
  7. Usability and Interface Simplicity
  8. Architectural Adaptability
  9. Performance Benchmarking
  10. Conclusion
  11. Implementation Details
  12. Code Example for a Modular Pipeline
  13. Standardized Interface for Integrating Customized Components

Figures (1)

  • Figure 1: Modular abstraction of pipeline construction using FMTK: The framework allows instantiating pipelines by pairing an FM with interchangeable components. Users can dynamically select and load components, specify trainable parts (e.g., decoder), and benchmark pipelines in a unified interface. We illustrate two example configurations using the same FM: (top)encoder-decoder-adapter-tuned pipeline with E1, A1 and D2; (bottom) encoder-decoder tuned with E3 and D3.