Skip to content

Repository files navigation

DeepTempo

Fine-tune small time-series foundation models on your own data. Rough edges included.

License: Apache 2.0 Python 3.10+ Tests

Status: early-stage, under active development. Not production-ready. Expect breaking changes.

What is this?

DeepTempo is a toolkit for fine-tuning pre-trained time-series transformer models on domain-specific forecasting tasks. The goal is to lower the barrier for practitioners who want to adapt foundation models to their own data without writing everything from scratch.

Think of it as "HuggingFace Trainer meets time-series" — a narrow, focused tool, not a platform.

What this is NOT

  • A model zoo with hundreds of pre-trained weights
  • A SaaS or managed service
  • Production-hardened infrastructure
  • A research framework for designing new architectures

Quick Start

Install

About

Time-series foundation model fine-tuning toolkit with GPU acceleration

Topics

Resources

Contributing

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages