Contamination MapTime series / Model
Granite-TimeSeries-TTM-R2
- Organisation
- IBM Research
- Family
- Tiny Time Mixers (TTM) · version R2
- Released
- 2024-10-08 (HF repo creation (createdAt) of ibm-granite/granite-timeseries-ttm-r2 (first weights commit '512-96-r2' on 2024-10-09), https://huggingface.co/api/models/ibm-granite/granite-timeseries-ttm-r2)
- Checkpoints
- ibm-granite/granite-timeseries-ttm-r2, ibm-granite/granite-timeseries-ttm-r2@main, ibm-granite/granite-timeseries-ttm-r2@512-192-r2, ibm-granite/granite-timeseries-ttm-r2@512-336-r2, ibm-granite/granite-timeseries-ttm-r2@512-720-r2, ibm-granite/granite-timeseries-ttm-r2@1024-96-r2, ibm-granite/granite-timeseries-ttm-r2@1024-192-r2, ibm-granite/granite-timeseries-ttm-r2@1024-336-r2, ibm-granite/granite-timeseries-ttm-r2@1024-720-r2, ibm-granite/granite-timeseries-ttm-r2@1536-96-r2, ibm-granite/granite-timeseries-ttm-r2@1536-192-r2, ibm-granite/granite-timeseries-ttm-r2@1536-336-r2, ibm-granite/granite-timeseries-ttm-r2@1536-720-r2
- Weights
- Open · apache-2.0
- Links
- PaperModel cardCode
Training data disclosure: Full list
Level determined for 100% of benchmark cells.
- TestTest data seen0
- SeenSeen, split unclear4
- TrainTraining data seen0
- Related data seen0
- NoNot seen45
- ?Unknown0
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Other declared training data
-
Australian Electricity Demand · split: unknown
Evidence
“Australian Electricity Demand: https://zenodo.org/records/4659727”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Australian Weather · split: unknown
Zenodo 4654822 is the Monash 'Weather Dataset' (Australian daily weather), not the long-horizon Weather dataset.
Evidence
“Australian Weather: https://zenodo.org/records/4654822”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Bitcoin · split: unknown
Evidence
“Bitcoin: https://zenodo.org/records/5122101”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
KDD Cup 2018 · split: unknown
Evidence
“KDD Cup 2018: https://zenodo.org/records/4656756”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
London Smart Meters · split: unknown
Evidence
“London Smart Meters: https://zenodo.org/records/4656091”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Saugeen River Flow · split: unknown
Evidence
“Saugeen River Flow: https://zenodo.org/records/4656058”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Solar Power · split: unknown
Evidence
“Solar Power: https://zenodo.org/records/4656027”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Sunspots · split: unknown
Evidence
“Sunspots: https://zenodo.org/records/4654722”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Solar (LSTNet, 137 Alabama PV plants) · split: unknown
Matched via the cited Zenodo record 4656144 (Monash Solar 10 Minutes, the 137 Alabama plants of Lai et al.).
Evidence
“Solar: https://zenodo.org/records/4656144”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
US Births · split: unknown
Evidence
“US Births: https://zenodo.org/records/4656049”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Wind Farms Production · split: unknown
Evidence
“Wind Farms Production: https://zenodo.org/records/4654858”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
Wind Power · split: unknown
Evidence
“Wind Power: https://zenodo.org/records/4656032”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
PEMSD3 · split: unknown
Evidence
“PEMSD3, PEMSD4, PEMSD7, PEMSD8, PEMS_BAY: https://drive.google.com/drive/folders/1g5v2Gq1tkOq8XO0HDCZ9nOTtRpB6-gPe”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
PEMSD4 · split: unknown
Evidence
“PEMSD3, PEMSD4, PEMSD7, PEMSD8, PEMS_BAY: https://drive.google.com/drive/folders/1g5v2Gq1tkOq8XO0HDCZ9nOTtRpB6-gPe”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
PEMSD7 · split: unknown
Evidence
“PEMSD3, PEMSD4, PEMSD7, PEMSD8, PEMS_BAY: https://drive.google.com/drive/folders/1g5v2Gq1tkOq8XO0HDCZ9nOTtRpB6-gPe”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
PEMSD8 · split: unknown
Evidence
“PEMSD3, PEMSD4, PEMSD7, PEMSD8, PEMS_BAY: https://drive.google.com/drive/folders/1g5v2Gq1tkOq8XO0HDCZ9nOTtRpB6-gPe”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
PEMS_BAY · split: unknown
Evidence
“PEMSD3, PEMSD4, PEMSD7, PEMSD8, PEMS_BAY: https://drive.google.com/drive/folders/1g5v2Gq1tkOq8XO0HDCZ9nOTtRpB6-gPe”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15 -
LOS_LOOP · split: unknown
Evidence
“LOS_LOOP: https://drive.google.com/drive/folders/1g5v2Gq1tkOq8XO0HDCZ9nOTtRpB6-gPe”
huggingface.co · model card · Training Data section (r2 list) · retrieved 2026-10-15
Declared exclusions
-
Dataset Loop Seattle (LibCity) · entirely excluded
Evidence
“Please note that, the last three datasets in the Libcity section have been excluded from the pre-training process for the model releases intended for enterprise-use.”
arxiv.org · paper · Appendix C.1 / Table 8 caption · retrieved 2026-10-15 -
Dataset SZ-Taxi (LibCity) · entirely excluded
Evidence
“Please note that, the last three datasets in the Libcity section have been excluded from the pre-training process for the model releases intended for enterprise-use.”
arxiv.org · paper · Appendix C.1 / Table 8 caption · retrieved 2026-10-15
Benchmark cells
GIFT-Eval
Not seen Jena Weather (GIFT-Eval) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen BizITObs - Application · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen BizITObs - Service · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen BizITObs - L2C · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Bitbrains - Fast Storage · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Bitbrains - rnd · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Restaurant (Recruit Restaurant Visitor Forecasting) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen ETT1 (GIFT-Eval) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen ETT2 (GIFT-Eval) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Loop Seattle (LibCity) · the authors state the dataset was excluded from training
Evidence (1)
“Please note that, the last three datasets in the Libcity section have been excluded from the pre-training process for the model releases intended for enterprise-use.”
Not seen SZ-Taxi (LibCity) · the authors state the dataset was excluded from training
Evidence (1)
“Please note that, the last three datasets in the Libcity section have been excluded from the pre-training process for the model releases intended for enterprise-use.”
Not seen M_DENSE (LibCity) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Seen, split unclear Solar (LSTNet, 137 Alabama PV plants) · the dataset was used; the source does not say which part
Evidence (1)
“Solar: https://zenodo.org/records/4656144”
Not seen Hierarchical Sales (Mancuso et al., 2020) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen M4 Yearly · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen M4 Quarterly · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen M4 Monthly · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen M4 Weekly · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen M4 Daily · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen M4 Hourly · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Hospital · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen COVID Deaths · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Seen, split unclear US Births · the dataset was used; the source does not say which part
Evidence (1)
“US Births: https://zenodo.org/records/4656049”
Seen, split unclear Saugeen River Flow · the dataset was used; the source does not say which part
Evidence (1)
“Saugeen River Flow: https://zenodo.org/records/4656058”
Not seen Temperature Rain · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Seen, split unclear KDD Cup 2018 · the dataset was used; the source does not say which part
Evidence (1)
“KDD Cup 2018: https://zenodo.org/records/4656756”
Not seen Car Parts · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen ElectricityLoadDiagrams20112014 (UCI) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Long-horizon forecasting (ETT, Electricity, Traffic, Weather, Exchange, ILI)
Not seen ETTh1 · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen ETTh2 · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen ETTm1 · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen ETTm2 · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Electricity (ECL, 321 clients) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Traffic (PeMS San Francisco Bay, 862 sensors) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Weather (Jena, 2020; Autoformer) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Exchange-Rate · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen ILI (CDC influenza-like illness, national) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Time-series anomaly detection
Not seen TSB-AD-U (TSB-AD univariate) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen TSB-AD-M (TSB-AD multivariate) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen UCR Time Series Anomaly Archive (Hexagon ML/UCR, KDD Cup 2021) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Server Machine Dataset (SMD) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Mars Science Laboratory (MSL) telemetry anomaly dataset · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Soil Moisture Active Passive (SMAP) telemetry anomaly dataset · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Pooled Server Metrics (PSM) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Secure Water Treatment (SWaT) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Yahoo S5 Labeled Anomaly Detection Dataset (Yahoo Webscope S5) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen Numenta Anomaly Benchmark (NAB) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Time-series classification
Not seen UCR Time Series Classification Archive (2018) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Not seen UEA Multivariate Time Series Classification Archive (2018) · absent from the complete list of training data
Evidence (2)
“The r2 TTM models were trained on a collection of datasets as follows:”
“TTM r2 comprises TTM variants pre-trained on larger pretraining datasets (\~700M samples).”
Notes
Branches without the r2.1 suffix. Model card: pre-training data 'differs slightly' from the paper. The listed datasets are a subset of paper Table 8 (enterprise release drops LOOP_SEATTLE, SZ_TAXI, Q-TRAFFIC; also kaggle_web_traffic and nn5_daily from Table 8 do not appear in the r2 list). The paper's Weather/Electricity/Traffic/ETT exclusion is stated for the paper's pretraining set, not restated in this card. GIFT-Eval leaderboard configs for TTM-R2-Pretrained/Finetuned mark "testdata_leakage": "Yes" (the leaderboard notebook used TTM-r2 models from this repo, possibly r2.1 branches). Other stated exclusions, not tracked here: Q-TRAFFIC (all).
Sources
- Granite-TimeSeries-TTM-R2 model card · model card · retrieved 2026-10-15
- Tiny Time Mixers (TTMs) (arXiv v8) · paper · retrieved 2026-10-15
- GIFT-Eval leaderboard config for TTM-R2-Pretrained · leaderboard · retrieved 2026-10-15
- GIFT-Eval leaderboard config for TTM-R2-Finetuned · leaderboard · retrieved 2026-10-15