Skip to content

[Feature Request]: TiRex in Darts #29

@daidahao

Description

@daidahao

Describe the feature request

Dear @martinloretzzz and NXAI team,

Thank you for your work on TiRex. I am truly impressed by its performance on GIFT-EVAL and fev. For the past few months, I've been a key contributor to Darts, a popular time series library, and so far provided latest foundation models such as Chronos-2 and TimesFM 2.5 under sklearn-like forecasting APIs.

Naturally, it would be remiss of us to not include TiRex in Darts considering its strong performance and efficiency. Therefore, would you be open to contributing a TiRex model implementation in Darts? I am happy to assist best I can or co-author a PR should you prefer.

In terms of technicality, Darts offers two ways of implementing foundation models and you are free to chose either of them:

  • GlobalForecastingModel which is a base class to interface external libraries such as tirex-ts.
  • FoundationModel which is a base class to allow a PyTorch-native implementation and uses HuggingFaceConnector to load model weights from HuggingFace Hub, same way as tirex-ts does.

Let me know what you think of the proposal. Thank you and your team for open-sourcing TiRex.

Best,
Zhihao

Describe scenario use case

No response

Additional Context

No response

Priority

Medium

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions