README.md

metadata

pipeline_tag: time-series-forecasting
library_name: timestral
tags:
  - time-series
  - forecasting
  - foundation-model
model-index:
  - name: Timestral-28M
    results:
      - task:
          type: time-series-forecasting
          name: Time Series Forecasting
        dataset:
          name: GIFT-Eval
          type: Salesforce/GiftEval
        metrics:
          - name: MASE
            type: mase
            value: 0.678
          - name: CRPS
            type: crps
            value: 0.461
        source:
          name: GIFT-Eval leaderboard
          url: https://huggingface.co/spaces/Salesforce/GIFT-Eval
      - task:
          type: time-series-forecasting
          name: Time Series Forecasting
        dataset:
          name: TIME
          type: Real-TSF/TIME
        metrics:
          - name: MASE
            type: mase
            value: 0.638
          - name: CRPS
            type: crps
            value: 0.534
        source:
          name: TIME leaderboard
          url: https://huggingface.co/spaces/Real-TSF/TIME-leaderboard
      - task:
          type: time-series-forecasting
          name: Time Series Forecasting
        dataset:
          name: fev-bench
          type: autogluon/fev-bench
        metrics:
          - name: SQL skill score
            type: skill-score
            value: 48.42
          - name: MASE skill score
            type: skill-score
            value: 37.27
        source:
          name: fev-bench leaderboard
          url: https://huggingface.co/spaces/autogluon/fev-bench
license: other

Timestral-28M

Timestral is a family of time series foundation models for zero-shot forecasting, available in 5M, 28M, 123M, and 209M parameter versions. This repository contains the 28M checkpoint.

For more information about our Timestral family, go to the main model page: Timestral-209M.

Model Details

• Size: 28M parameters.
• Architecture: patch-based transformer with time attention and variate attention, 12 layers, embedding dimension 384.
• Patching: 32 time steps per patch. Each patch is embedded from its values together with their missing-value indicators.
• Context length: up to 32,768 time steps (1,024 patches).
• Forecast horizon: Up to 1,024 time steps (32 patches) in a single forward pass. Longer horizons are forecast block by block, with each block's median forecast appended to the context.
• Outputs: 21 quantiles (0.01, 0.05, 0.1, 0.15, …, 0.85, 0.9, 0.95, 0.99).
• Inputs: univariate or multivariate target variates, plus past-only and known-future covariates, numeric or categorical (categorical covariates are target-encoded). Missing values are marked as NaN.
• Inference: PyTorch on CPU or GPU.

Benchmarks

Timestral-209M achieves state-of-the-art zero-shot accuracy across three major forecasting benchmarks: GIFT-Eval, TIME, and fev-bench.

Smaller Timestral models offer a trade-off between model size and forecasting performance. All Timestreal models outperform every other model at the respective model size. See the Pareto-front plot on the main model page.

Results across the Timestral family

Benchmark	Metric	Timestral-5M	Timestral-28M	Timestral-123M	Timestral-209M
GIFT-Eval	MASE ↓	0.705	0.678	0.659	0.658
	CRPS ↓	0.479	0.461	0.448	0.446
TIME	MASE ↓	0.652	0.638	0.628	0.626
	CRPS ↓	0.545	0.534	0.525	0.522
fev-bench	SQL skill score ↑	46.19	48.42	50.18	50.39
	MASE skill score ↑	34.54	37.27	39.19	39.49

License

Models have been released under a Modified MIT license: See here.

You must not use this model in a manner that infringes, misappropriates, or otherwise violates any third party’s rights, including intellectual property rights