Contamination Map Methodology

Time series / Model

UniTS

Organisation
Harvard University (MIMS lab)
Family
UniTS
Released
2024-02-29 (arXiv v1 submission date (https://arxiv.org/abs/2403.00131v1) and GitHub release 'ckpt' published 2024-02-29 (https://github.com/mims-harvard/UniTS/releases/tag/ckpt))
Checkpoints
units_x128_pretrain_checkpoint.pth, units_x32_pretrain_checkpoint.pth, units_x64_supervised_checkpoint.pth
Weights
Open · MIT
Links
PaperCode

Training data disclosure: Full list

Level determined for 100% of benchmark cells.

“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Other declared training data

Benchmark cells

GIFT-Eval

Related data seen Jena Weather (GIFT-Eval) · related dataset Weather (Jena, 2020; Autoformer) was used

Evidence (6)
“WeatherP 96 [110]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15
“We incorporate Jena Weather5 dataset following Autoformer (Wu et al., 2021).”
arxiv.org · paper · Appendix D (footnote 5 points to https://www.bgc-jena.mpg.de/wetter/) · retrieved 2026-10-15
“is recorded every 10 minutes for 2020 whole year, which contains 21 meteorological indicators, such as air temperature, humidity, etc.”
arxiv.org · paper · Section 4 (Datasets) · retrieved 2026-10-15
“https://www.bgc-jena.mpg.de/wetter/”
arxiv.org · paper · Section 4, footnote 3 · retrieved 2026-10-15

Not seen BizITObs - Application · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen BizITObs - Service · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen BizITObs - L2C · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Bitbrains - Fast Storage · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Bitbrains - rnd · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Restaurant (Recruit Restaurant Visitor Forecasting) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Related data seen ETT1 (GIFT-Eval) · related dataset ETTh1 was used

Evidence (5)
“ETTh1P 96 [127]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15
“From Informer (Zhou et al., 2021) we utilize ETT1 and ETT2 datasets, which denote electricity transformer temperature”
arxiv.org · paper · Appendix D · retrieved 2026-10-15
“ETTh1&ETTh2 7 17,420”
arxiv.org · paper · Table 1 · retrieved 2026-10-15

Not seen ETT2 (GIFT-Eval) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Loop Seattle (LibCity) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen SZ-Taxi (LibCity) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen M_DENSE (LibCity) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Solar (LSTNet, 137 Alabama PV plants) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Hierarchical Sales (Mancuso et al., 2020) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen M4 Yearly · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen M4 Quarterly · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen M4 Monthly · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen M4 Weekly · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen M4 Daily · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen M4 Hourly · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Hospital · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen COVID Deaths · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen US Births · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Saugeen River Flow · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Temperature Rain · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen KDD Cup 2018 · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Car Parts · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Related data seen ElectricityLoadDiagrams20112014 (UCI) · related dataset Electricity (ECL, 321 clients) was used

Evidence (6)
“ECLP 96 [102]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15
“The raw dataset is in https://archive.ics.uci.edu/ml/datasets/ElectricityLoadDiagrams20112014.”
raw.githubusercontent.com · readme · Electricity consumption · retrieved 2026-10-15
“Because the some dimensions are equal to 0. So we eliminate the records in 2011. Final we get data contains electircity consumption of 321 clients from 2012 to 2014.”
raw.githubusercontent.com · readme · Electricity consumption · retrieved 2026-10-15
“https://archive.ics.uci.edu/ml/datasets/ElectricityLoadDiagrams20112014”
arxiv.org · paper · Section 4, footnote 1 · retrieved 2026-10-15

Long-horizon forecasting (ETT, Electricity, Traffic, Weather, Exchange, ILI)

Training data seen ETTh1 · only the training split was used, not the benchmark's test data

Evidence (3)
“ETTh1P 96 [127]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Not seen ETTh2 · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Related data seen ETTm1 · related dataset ETTh1 was used

Evidence (4)
“ETTh1P 96 [127]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15
“we also provide the hourly-level variants for fast development (marked by *h*), i.e. ETT-small-h1 and ETT-small-h2.”
raw.githubusercontent.com · readme · ETT-small · retrieved 2026-10-15

Not seen ETTm2 · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Training data seen Electricity (ECL, 321 clients) · only the training split was used, not the benchmark's test data

Evidence (3)
“ECLP 96 [102]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Training data seen Traffic (PeMS San Francisco Bay, 862 sensors) · only the training split was used, not the benchmark's test data

Evidence (3)
“TrafficP 96 [85]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Training data seen Weather (Jena, 2020; Autoformer) · only the training split was used, not the benchmark's test data

Evidence (3)
“WeatherP 96 [110]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Training data seen Exchange-Rate · only the training split was used, not the benchmark's test data

Evidence (3)
“ExchangeP 192 [53]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Training data seen ILI (CDC influenza-like illness, national) · only the training split was used, not the benchmark's test data

Evidence (3)
“ILIP 60 [11]”
arxiv.org · paper · Table 7 (forecasting rows) · retrieved 2026-10-15
“For subsets of a dataset such as ETTh1, we start by splitting the data into training and testing sets based on distinct time intervals of a long time series sequence, following splits in [112].”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Time-series anomaly detection

Not seen TSB-AD-U (TSB-AD univariate) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen TSB-AD-M (TSB-AD multivariate) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen UCR Time Series Anomaly Archive (Hexagon ML/UCR, KDD Cup 2021) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Server Machine Dataset (SMD) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Mars Science Laboratory (MSL) telemetry anomaly dataset · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Soil Moisture Active Passive (SMAP) telemetry anomaly dataset · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Pooled Server Metrics (PSM) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Secure Water Treatment (SWaT) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Yahoo S5 Labeled Anomaly Detection Dataset (Yahoo Webscope S5) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Not seen Numenta Anomaly Benchmark (NAB) · absent from the complete list of training data

Evidence (3)
“For multi-task learning on forecasting and classification, we compiled 38 datasets from several sources [79, 33, 82].”
arxiv.org · paper · Section 5 · retrieved 2026-10-15
“Table 7: Multi-task datasets for classification and forecasting.”
arxiv.org · paper · Table 7 caption · retrieved 2026-10-15
“For transfer learning on new tasks and datasets, we load the model weights pre-trained on 38 datasets”
arxiv.org · paper · Section 5 · retrieved 2026-10-15

Time-series classification

Training data seen UCR Time Series Classification Archive (2018) · only the training split was used, not the benchmark's test data

Evidence (3)
“SharePriceIncrease [79] JapaneseVowels [52] SpokenArabicDigits [5] Heartbeat [61] ECG5000 [35] NonInvasiveFetalECGThorax1 [95] Blink [19] FaceDetection [40] SelfRegulationSCP2 [7] ElectricDevices [60] Trace [93] FordB [23] MotionSenseHAR [75] EMOPain [27] UWaveGestureLibrary [63] Chinatown [23] MelbournePedestrian [23] PEMS-SF [20]”
arxiv.org · paper · Table 7 (classification rows) · retrieved 2026-10-15
“Datasets were aggregated from the Monash Forecasting Repository [33], Time Series Classification Website [79], and Time Series Library [112]. The combined training set consists of over 35 million timesteps and over 6,000 variables.”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Training data seen UEA Multivariate Time Series Classification Archive (2018) · only the training split was used, not the benchmark's test data

Evidence (3)
“SharePriceIncrease [79] JapaneseVowels [52] SpokenArabicDigits [5] Heartbeat [61] ECG5000 [35] NonInvasiveFetalECGThorax1 [95] Blink [19] FaceDetection [40] SelfRegulationSCP2 [7] ElectricDevices [60] Trace [93] FordB [23] MotionSenseHAR [75] EMOPain [27] UWaveGestureLibrary [63] Chinatown [23] MelbournePedestrian [23] PEMS-SF [20]”
arxiv.org · paper · Table 7 (classification rows) · retrieved 2026-10-15
“Datasets were aggregated from the Monash Forecasting Repository [33], Time Series Classification Website [79], and Time Series Library [112]. The combined training set consists of over 35 million timesteps and over 6,000 variables.”
arxiv.org · paper · Appendix (Datasets) · retrieved 2026-10-15
“_, train_loader_list = self._get_data(flag='train')”
raw.githubusercontent.com · code · exp/exp_pretrain.py · retrieved 2026-10-15

Notes

Self-supervised (x32/x128) and multi-task supervised (x64) checkpoints, all trained on the training splits of the 38 Table 7 tasks (20 forecasting tasks over 7 datasets + 18 classification datasets). Classification members: 7 UCR and 7 UEA datasets. No anomaly data in pre-training; anomaly detection is few-shot fine-tuning (see units-few-shot-anomaly-detection). Other released fine-tuned checkpoints (few-shot new-data classification/forecasting incl. ECG200, SelfRegulationSCP1, RacketSports, Handwriting, Epilepsy, StarLightCurves from Table 8; imputation) are not separate entries here.

Sources