Repository Issue Activity (beta)

#3196 Add support for Google's TimesFM 3.0 foundation model (multivariate, past & future covariates)

#1618 Integrate darts models with MLflow (logging, autologging, loading)

#3179 Support cold-start forecasting for Torch models without target history

#3177 Safe-by-default model loading

#3128 [QUESTION] Have you considered more advanced outlier detection and value imputation?

#1321 Raise an appropriate exception in case of HTTP 404 when downloading datasets. (Currently a CSV containing "404 not found" is created, which raises a hash-check exception)

#2842 Add One-Hot Encoder to Time Axis Encoders

#3171 Allow TFTExplainer to explain more series than the model batch size

#2685 Add support for TFT feature importance over time

#2489 TFTExplainer: getting explanations when trained on multiple time series

#3001 [BUG] AutoETS fails on short time series

#2758 [FEAT] RIN for covariates

#1649 Retrieve last hidden state for RNNModel and BlockRNNModel

#3132 Consistent zero-denominator handling for percentage and range metrics

#3008 [BUG] TFTModel creation doesn't raise exception from model.fit_from_dataset and empty future covariates

#2638 [BUG] TCN model cannot be saved when used with callbacks

#2043 future_covariate support to NBEATS & NHits models

#3156 [BUG] Warnings regarding different input and output data types even if they are the same when loading models from checkpoint

#2968 multi_models = True and output_chunk_length > 1 : index future covariates relatively to the forecasted step

#3016 [New Model] TiRex Foundation Model

#1845 Add a multivariate wrapper around univariate models

#3150 [BUG] WindowsTransfomer treat_na not working

#672 [BUG] Implementation errors on `TransformerModel`.

#601 Refactor Transformer model

#3148 Segmentation fault in KalmanForecaster with Python 3.13, numpy 2.4+, pandas 3.0+ on Linux