NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
Probabilistic time series modeling in Python.
Last release 2 months ago
31 Jul 2026
Release timing varies
gaps range from 8 days to 12 months
Nearly every release is documented
notes for 60 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
7 years old
114 releases · first in 2019
Drop Python 3.8 from workflows, run on 3.11 by @lostella in https://github.com/awslabs/gluonts/pull/3228
observed=True inside DataFrame.groupby in PandasDataset to silence the FutureWarning by @shchur in https://github.com/awslabs/gluonts/pull/3254<2.5 by @abdulfatir in https://github.com/awslabs/gluonts/pull/3281Full Changelog: https://github.com/awslabs/gluonts/compare/v0.16.0...v0.17.0
One column per quarter.
observed=True inside DataFrame.groupby in PandasDataset to silence the FutureWarning by @shchur in #3254<2.5 by @abdulfatir in #3281Full Changelog: v0.16.0...v0.17.0
Backport: Use --no-project in pypi workflow
Backport: Use --no-project in pypi workflow (#3328)
*Description of changes:* Backport of #3327
By submitting this pull request, I confirm that you can use, modify,
copy, and redistribute this contribution, under the terms of your
choice.
**Please tag this pr with at least one of these labels to make our
release process faster:** BREAKING, new feature, bug fix, other change,
dev setup
Update numpy upper bound to <2.5 #3281
Backported fixes:
Full Changelog: https://github.com/awslabs/gluonts/compare/v0.16.2...v0.16.3
Backported fixes:
Full Changelog: v0.16.2...v0.16.3
Make observed_value_field optional in TFTInstanceSplitter #3259 by @abdulfatir
Full Changelog: https://github.com/awslabs/gluonts/compare/v0.16.1...v0.16.2
Full Changelog: v0.16.1...v0.16.2
Allow disabling scaler for DL models #3251 by @shchur
Full Changelog: https://github.com/awslabs/gluonts/compare/v0.16.0...v0.16.1
Full Changelog: v0.16.0...v0.16.1
PatchTST: Add support for dynamic features by @rshyamsundar in https://github.com/awslabs/gluonts/pull/3167
Full Changelog: https://github.com/awslabs/gluonts/compare/v0.15.1...v0.16.0
Full Changelog: v0.15.1...v0.16.0
Docs: fix custom pytorch model tutorial by @lostella in https://github.com/awslabs/gluonts/pull/3188
<!--
Full Changelog: https://github.com/awslabs/gluonts/compare/v0.15.1...v0.16.0rc1 -->
Fix incorrect input routing for models #3186 (by @shchur)
Backporting fixes:
add iTransformer multivariate forecaster (#3017) by @kashif
Fix type annotation for device in PyTorchPredictor #3094 by @lostella
Backporting fixes:
Fix Rotbaum serialization and deserialization #3068 by @pantanurag555
Backporting:
Fix iterable.Cached. #3060 by @jaheba
Backporting fixes:
Fix edge cases in metric computation #3037 (@lostella)
Backporting fixes:
See full diff here: https://github.com/awslabs/gluonts/compare/v0.13.0...v0.14.0
See full diff here: https://github.com/awslabs/gluonts/compare/v0.13.0...v0.14.0
Nothing published for this version
Nothing published for this version
Fix Rotbaum serialization and deserialization #3068 by @pantanurag555
Backporting:
Fix edge cases in metric computation #3037 (@lostella)
Backporting fixes:
Fixing issues with updated requirements, which would prevent the package from installing alongside old versions of PyTorch Lightning:
Fixing issues with updated requirements, which would prevent the package from installing alongside old versions of PyTorch Lightning:
[torch] Return a model even if callback has no best model path #2952 by @kashif
Backporting fixes:
Raise warning in QuantileForecast.mean when mean is not there #2843 by @lostella
Backporting fixes:
Turn type comparison into isinstance #2958 by @lostella
Backporting fixes:
Zebras: Fix index handling of SplitFrame.resize. #2938 by @jaheba
Backporting fixes:
Fix NaNs in seasonal error #2894 by @gorold
Backporting fixes:
Speedup is_uniform for PandasDataset. #2878 by @jaheba
Backporting fixes:
We're happy to release gluonts version 0.13! This release contains a few new features and breaking changes compared to 0.12, especially around PyTorch…
We're happy to release gluonts version 0.13! This release contains a few new features and breaking changes compared to 0.12, especially around PyTorch models, data handling and model evaluation:
There are several more improvements and fixes compared to 0.12, which you can find in the changelog below. This release was possible thanks to the great work of several contributors: @jaheba, @MarcelK1102, @lostella, @gorold, @kashif, @dcmaddix, @abdulfatir, @melopeo, @huibinshen, @shchur, @pablovicente, @Gandor26, @Linbo-Liu. Thanks everyone, and thanks to users and authors of issue reports for the precious feedback!
Nothing published for this version
Remove .to_timestamp() to fix interval plotting #2800 by @abdulfatir
Backporting fixes:
Test: Set caplog level for shell tests. #2786 by @jaheba
Backporting fixes:
Fix pandas removed deprecations in tests #2778 by @lostella
Backporting fixes:
Allow PyTorch 2.0 #2724 by @lostella
Backporting fixes:
Fix ev.seasonal_error #2696 by @lostella
Backporting fixes:
Delay instantiation of ScipyStudentT object #2660 by @gorold
Backporting fixes:
Fix PyTorch training loop #2643 by @gorold
Backporting fixes:
Fix: torch PoissonOutput scaling #2619 by @kashif
Backporting fixes:
Support for Python 3.6 is dropped (#2542).
Support for Python 3.6 is dropped (#2542).
Models:
statsforecast models (#2360, #2515, #2561)gluonts.ext (#2362, #2597)Data:
item_id, and the dtype of each column determinins
which are numerical vs categorical features, with automated detection of cardinalities
in the latter case. See the updated tutorial notebook on how to use it.Evaluation:
gluonts.ev (#2450) will gradually replace the existing
gluonts.evaluation as an improved, more flexible alternative.weight_decay in torch TFT estimator class (#2603) @goroldPandasDataset (#2599) @lostellagluonts.util.safe_extract (#2606) @lostellatsbench, apply latest black (#2613) @lostellagluonts.model (#2597) @lostellaNothing published for this version
Fix _version location for sdist. #2729 by @jaheba
Backporting fixes:
Faster index building in PandasDataset #2663 by @huibinshen
Backporting fixes:
Fix PyTorch training loop #2643
Backporting fixes:
Fix requirements following breaking change in setuptools #2604 by @lostella
Backporting fixes:
Update workflow actions to latest versions #2447 by @jaheba
Backporting fixes:
Make serde.dataclass always kw-only. (#2428 by @jaheba)
Backporting fixes:
serde.dataclass inheritance handling. (#2512 by @jaheba)QuantileForecast.quantile in case only mean is stored (#2513 by @lostella)aggregate_valid for non-numerical columns (#2526 by @lostella)itertools.select. #2426 by @jaheba
Backporting fixes:
Backports for v0.11.5. by @jaheba in https://github.com/awslabs/gluonts/pull/2491
Full Changelog: https://github.com/awslabs/gluonts/compare/v0.11.4...v0.11.5
Fix pandas issue with inferring start of X frequency. (#2462 by @jaheba)
Backports:
Add test cases for PandasDataset, fix missing assertion (#2453 by @lostella)
Backporting fixes:
Add test cases for PandasDataset, fix missing assertion (#2453 by @lostella)
Backporting fixes:
Fix rotbaum random seed and num_samples argument. (#2408 by @sighellan)
Backporting fixes:
Fix dominick dataset bug. (#2364 by @haskarb)
Backporting fixes:
Docs: fix tutorial after breaking changes in trainer class (#2179 by @lostella)
Estimators are now re-trainable on new data, using the train_from method. This accepts a previously trained model (predictor), and new data to train on, and can greatly reduce training time if combined with early stopping. The feature is integrated with gluonts.shell-based SageMaker containers, and can be used by specifying the additional model channel to point to the output of a previous training job. More info in #2249.
Two models are added in this release:
DeepVARHierarchicalEstimator, a hierarchical extension to DeepVAREstimator; learn more about how to use this in this tutorial.DeepNPTSEstimator, a global extension to NPTS, where sampling probabilities are learned from data; learn more on how to use this estimator here.This release moves MXNet-based models from gluonts.model to gluonts.mx.model; the old import paths continue working in this release, but are deprecated and will be removed in the next release. For example, now the MXNet-based DeepAREstimator should be imported from gluonts.mx (or gluonts.mx.model.deepar).
We also removed deprecated options for learning rate reduction in the gluonts.mx.Trainer class: these can now be controlled via the LearningRateReduction callback.
We updated the functionality to split time series datasets (along the time axis) for training/validation/test purposes. Now this functionality can be easily accessed via the split function (from gluonts.dataset.split import split); learn more about this here.
This feature is experimental and subject to future changes.
(input, output) pairs in the test split (#2031 by @npnv)mx.model. (#2126 by @Hongqing-work)mx.Trainer. (#2153 by @jaheba)gluonts.dataset.split code, test, docs (#2223 by @lostella)mx.Trainer stops training. (#2131 by @Hongqing-work)serde.dataclass. (#2166 by @jaheba)dataset.schema.translate. (#2304 by @jaheba)forecast_start to entry-wise metrics in evaluator (#2312 by @lostella)PandasDataset for Python 3.9 (#2141 by @lostella)PandasDataset faster (#2148 by @lostella)Timestamp instead of Period (#2182 by @lostella)gluonts.mx.batchify.stack (#2184 by @lostella)item_id values in ConstantValuePredictor (#2192 by @codingWhale13)SymbolBlock serde issue (#2187 by @lostella)r_forecast wrapper to shift start date when truncating time series (#2216 by @abdulfatir)to_quantile_forecast (#2225 by @eugeneteoh)gluonts.mx.trainer.Trainer in case of empty data loader (#2228 by @lostella)TimeSeriesSlice performance (#2259 by @lostella)QuantileForecast.plot() to use DateTimeIndex (#2269 by @abdulfatir)gluonts.dataset.split for multivariate case (#2314 by @lostella)TestData class in gluonts.dataset.split (#2315 by @lostella)make_evaluation_predictions (#2309 by @lostella)kernel_size=1 (#2321 by @lostella)BinnedUniforms (#2344 by @moudheus)installation section. (#2130 by @jaheba)DataFrame example (#2150 by @lostella)holidays and matplotlib from core dependencies. (#2055 by @jaheba)Backporting fix: - Fix call to extractall #2648
Backporting fix:
Backports for v0.10.9 by @jaheba in https://github.com/awslabs/gluonts/pull/2610
Backporting fixes
Full Changelog: https://github.com/awslabs/gluonts/compare/v0.10.8...v0.10.9
Fix numerical bug in BinnedUniforms (#2344 by @moudheus)
Backporting fixes:
Add Github footer icon to docs. (#2285 by @jaheba)
Backporting fixes:
Improve len() for ParquetFile. (#2261 by @jaheba)
Backporting fixes:
Fix broken links in Available-models table (#2211 by @rshyamsundar)
Backporting fixes:
Full changelog: https://github.com/awslabs/gluon-ts/compare/v0.10.4...v0.10.5
Fix SymbolBlock serde issue (#2187 by @lostella)
Backporting fixes:
SymbolBlock serde issue (#2187 by @lostella)Fix Prophet wrapper to work with Timestamp instead of Period (#2182 by @lostella)
Backporting fixes:
Make PandasDataset faster (#2148 by @lostella)
Backport fixes:
PandasDataset faster (#2148 by @lostella)mx.Trainer stops training. (#2131 by @Hongqing-work)Your coding agent can read these notes before it upgrades. Set up the MCP server →