datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
TSFM-ScalingLaws-Dataset
TSFM-ScalingLaws-Dataset
This is the dataset for the paper Towards Neural Scaling Laws for Time Series Foundation Models.
Code: https://github.com/Qingrenn/TSFM-ScalingLaws
Well-trained models: https://huggingface.co/PeacefulData/TSFM-ScalingLaws-Checkpoints
Dataset Summary
Domain
Transport
Climate
Energy
Cloud
Health
Sales
Web
Total
Datasets
8
2
14
3
9
1
2
39
Time Points
4.82B
4.73B4.76B
2.15B
232M
140M
40M
16.8B
Proportion
28.52%
28.06%
28.21%
12.76%… See the full description on the dataset page: https://huggingface.co/datasets/Qingren/TSFM-ScalingLaws-Dataset.recursive-tsfm-data
Recursive TSFM training data
This repository contains the four source pools used for the experiments in
ecntu/recursive-tsfm, frozen as separate
Hugging Face datasets artifacts so sampling recipes remain explicit and auditable.
source
rows
origin
gifteval
500,000
Salesforce/GiftEvalPretrain
tsmixup
3,000,000
Chronos training_corpus_tsmixup_10m
kernel
600,000
Chronos training_corpus_kernel_synth_1m
gifteval_trainval
115,639
leakage-safe histories from… See the full description on the dataset page: https://huggingface.co/datasets/emiliocantuc/recursive-tsfm-data.p2-etf-tsfm-resultsgluco-tsfm-benchmark
Overview
This dataset aggregates multiple open-access CGM cohorts for blood glucose forecasting.
Included sub-datasets and their original download links
BIG_IDEA_LAB: https://doi.org/10.13026/zthx-5212
D1NAMO: https://zenodo.org/records/5651217
HUPA-UCM: https://data.mendeley.com/datasets/3hbcscwz44/1
Colas2019: https://pubmed.ncbi.nlm.nih.gov/31851681/
ShanghaiT2DM: https://doi.org/10.6084/m9.figshare.c.6310860
ShanghaiT1DM: https://doi.org/10.6084/m9.figshare.c.6310860… See the full description on the dataset page: https://huggingface.co/datasets/byluuu/gluco-tsfm-benchmark.real-v-tsfm
REAL-V-TSFM Dataset
REAL-V-TSFM is a novel time series dataset derived entirely from real-world video data using optical flow methods. It was created to evaluate the generalization capabilities of Time Series Foundation Models (TSFMs) on realistic temporal dynamics, bridging the gap between synthetic benchmarks and real data.
Dataset Overview
Extraction Method: Uses the Lucas-Kanade optical flow algorithm to track pixel trajectories at detected keypoints in… See the full description on the dataset page: https://huggingface.co/datasets/Volavion/real-v-tsfm.gluco_tsfm_benchmark
GlucoFM Benchmark - Aggregated CGM (TsFile)
Apache TsFile version of byluuu/gluco-tsfm-benchmark.
Overview
Blood-glucose forecasting benchmark aggregating open-access continuous
glucose monitoring (CGM) cohorts (11 datasets: BIG_IDEA_LAB, D1NAMO, HUPAD,
etc. - links in the source card). train and test each contain 529 subject
series; every source row is one subject's full CGM series with epoch-second
timestamps and glucose values in mg/dL (series lengths 26 ..… See the full description on the dataset page: https://huggingface.co/datasets/THULab/gluco_tsfm_benchmark.tsfm-peft-bench
TSFM-PEFT-Bench
A cross-architecture benchmark for evaluating Parameter-Efficient Fine-Tuning
(PEFT) recommendation reliability in Time Series Foundation Models (TSFMs).
Companion code and artifacts for the paper "TSFM-PEFT-Bench: A
Cross-Architecture Benchmark for PEFT Selection in Time Series Foundation
Models" (under double-blind review at NeurIPS 2026 Datasets and Benchmarks
Track).
Quick metadata:
License: Apache-2.0 (LICENSE)
Croissant manifest: tsfm_peft_bench.croissant.json… See the full description on the dataset page: https://huggingface.co/datasets/EvalData/tsfm-peft-bench.TSFMI
TSFMI-Synthetic
Synthetic time-series datasets with mathematically exact ground-truth labels
for the TSFMI baseline-controlled probing protocol.
Companion data for the NeurIPS 2026 Evaluations & Datasets Track submission
"TSFMI: A Baseline-Controlled Evaluation Protocol for Time-Series Foundation
Model Representations." The code (anonymous) lives at
https://anonymous.4open.science/r/TSFMI.
Why this dataset exists
Probing time-series foundation models (TSFMs) is hard… See the full description on the dataset page: https://huggingface.co/datasets/EvalData/TSFMI.future-ts
FUTURE-TS
Summary
FUTURE-TS v0.1.0 is a public-preview benchmark for time-series foundation
models. It treats evaluation as an executable protocol: task cards declare
issue times, horizons, available history, delayed labels, adaptation budgets,
resource limits, and metrics; submissions are validated, scored, and audited
against those contracts.
The core rule is future-only ranking: a submission is only ranked on labels
that were not available when the model was… See the full description on the dataset page: https://huggingface.co/datasets/TSFM-ai/future-ts.xacfed-tsfm-smart-grid-load-forecastingtsfm-har-benchrag_tsfmrag_tsfm_data