Update darts requirement from >=0.43 to >=0.44.0

[dependabot]

⚠️ Dependabot is rebasing this PR ⚠️

Rebasing might not happen immediately, so don't worry if this takes some time.

Note: if you make any changes to this PR yourself, they will take precedence over the rebase.

---

Updates the requirements on darts to permit the latest version.

Release notes

Sourced from darts's releases.

Darts minor 0.44.0

We are pleased to announce the release of a new Darts version.

You can find a list with all changes in the release notes.

Changelog

Sourced from darts's changelog.

0.44.0 (2026-04-30)

For users of the library:

Improved

• 🚀🚀 Dramatically reduced import times by deferring heavy third-party dependencies (torch, sklearn, scipy, ...) until they are actually needed. This benefits cold-start scenarios (serverless functions, CLI tools, short-lived scripts), CI pipelines, and interactive development workflows where fast feedback loops matter. Here are some import speed-up examples: #3066 by Dennis Bader
  • TimeSeries, metrics, datasets, data transformers: 8x faster (2.4 → 0.3 seconds)
  • Baseline models: 19x faster (5.7 → 0.3 seconds)
  • SKLearn models: 4.4x faster (5.7 → 1.3 seconds)
  • Torch models: 1.9x faster (5.7 → 3.0 seconds)
• 🚀🚀 Added new forecasting model TiRexModel : NX-AI's pre-trained 35M-parameter foundational model for zero-shot forecasting. It supports univariate, multivariate, and multiple time series forecasting without training and can output deterministic or probabilistic forecasts. #3038 by Lukas Fischer, Martin Loretz, and Zhihao Dai.
• Improvements to InvertibleDataTransformer and data Pipeline : #3085 by Jakub Chłapek
  • Added optional insample parameter to inverse_transform() to supply the transformed history needed by transformers like Diff, removing the need to manually concatenate the in-sample series before inverse-transforming a forecast.
  • Diff can now be properly used inside a target series Pipeline with historical_forecasts() via the data_transformers parameter. The backtested forecasts are correctly inverse-transformed back to the original scale.
• Improvements to RegressionEnsembleModel : #3041 by Gabriel Margaria and #3074 by Junghwan Na.
  • Base forecasting models using output_chunk_shift>0 are now fully supported. If you're using a custom regression_model, simply set its output shift to be the same as that of the base models.
  • Added support for output_chunk_length>1 for the ensemble (regression) model. This means that the ensemble model can now consume information from base model forecasts over the entire horizon.
• Improvements to Gradient-Boosted Models:
  • Added native multi-quantile support for CatBoostModel by using CatBoost’s MultiQuantile loss for faster training and inference. Set likelihood="multiquantile" to enable this feature. #3032 by Zhihao Dai
  • Added native multi-quantile support for XGBModel. Similar to the regular quantile support, it still fits dedicated models per quantile, but it is more efficient due to fewer tabularization operations. Set likelihood="multiquantile" to enable this feature. #3056 by Oswald Zink
• Improvements to metrics:
  • Added new probabilistic metrics crps() (Continuous Ranked Probability Score) and mcrps() (Mean CRPS): a proper scoring rule that generalises the Mean Absolute Error (MAE) to probabilistic forecasts. #3089 by Boubker Bennani
  • Added parameter name to all metric functions for customizing the displayed name. #3084 by Bruno Da Costa.
  • 🔴 Scaled metrics (ase, sse, mase, msse, rmsse) no longer raise a hard ValueError when the insample series has zero error scale (constant or perfectly seasonal signals). A new zero_division parameter controls the behavior: #3059 by Mahima Sharma
    • "warn" (default) raises a warning and returns np.nan for non-zero forecast errors, and 1.0 otherwise (forecast is on-par with naive forecast).
    • "raise" preserves the legacy error.
• Other improvements:
  • StatsForecastModel now accepts model as a StatsForecast model name, class, or instance; model_kwargs supplies constructor arguments when model is a name or class. This simplifies config-driven setups. #3058 by Trevin Chow.
  • 🔴 We moved NaiveEnsembleModel from darts.models.forecasting.baselines into a dedicated module darts.models.forecasting.naive_ensemble_model to separate the heavier dependencies from the baseline models and improve import times. The import remains identical from darts.models import NaiveEnsembleModel but pickled models that were saved in older Darts versions cannot be loaded anymore. To fix it, simply re-create the model and store it again. #3066 by Dennis Bader

Fixed

• Fixed a device mismatch error in TFTModel when moving a trained model to a different device (e.g., GPU to CPU for ONNX export). attention_mask and relative_index are now registered as non-persistent buffers so they are properly moved with the model. #3053 by Wolfhart Feldmeier
• Fixed several issues when using a StaticCovariatesTransformer with a OneHotEncoder (OHE) for categorical static covariates:
  • Fixed an issue where using the OHE parameter drop (e.g. drop="first", drop="if_binary") raised an exception due to wrong internal column mapping. #3065 by Jay Dasondee
  • Fixed an issue where using the OHE parameters min_frequency or max_categories silently truncated the column mapping. #3076 by Junghwan Na.
• Fixed a stale docstring in Diff that incorrectly stated a component_mask cannot be specified together with dropna=True. #3078 by Junghwan Na.

For developers of the library:

• Significantly reduced test suite session loading times by deferring heavy third-party dependencies until they are actually needed. #3072 by Dennis Bader
  • Test run sessions: 3x faster (6 → 2 seconds)
  • Test debug sessions: 6x faster (36 → 6 seconds)

0.43.0 (2026-03-23)

For users of the library:

Improved

... (truncated)

Commits

• d6c60a7 Bump version: 0.43.0 → 0.44.0
• d7d0a89 Release 0.44.0 (#3092)
• c0909e4 defer tirex import in tests (#3091)
• d3e693d Bump lxml-html-clean from 0.4.3 to 0.4.4 (#3090)
• c609740 Feature/tirex integration (#3038)
• 16fda02 Feat/metric crps (#3089)
• aa16909 Feat/insample transformers (#3085)
• 9d3c27a Feature/add name parameter to all metric functions (#3084)
• 415edfa Minor doc improvements / clarifications (#3082)
• 742358d fix: StaticCovariatesTransformer drops 'infrequent_sklearn' column with OneHo...
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)