AI Tools.

Search

time series forecasting by Salesforce

moirai-1.0-R-base

moirai-1.0-R-base is Salesforce's MOIRAI universal forecasting model, a transformer trained across a diverse mixture of time-series domains using the UNI2TS framework. It supports variable frequency (hourly, daily, weekly, etc.) and multivariate series with patch-based tokenization. The base variant is suitable for general-purpose zero-shot forecasting evaluation.

Summary text generated by an automated pipeline from the model card · Not individually reviewed or run by us · How this page is made

From the model card

Fields below are copied from the tags and counters on the HuggingFace repository Salesforce/moirai-1.0-R-base at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.

Publisher (HF namespace)
Salesforce
Pipeline tag
time-series-forecasting
Library
Transformers
Weight formats
safetensors
License tag
cc-by-nc-4.0 — read the license file in the repo before relying on it
Papers cited
arXiv:2402.02592
Downloads (HF counter at last fetch)
405,106
Likes (HF counter at last fetch)
32
Model card
https://huggingface.co/Salesforce/moirai-1.0-R-base

Use cases

  • Cross-domain zero-shot forecasting on new time-series datasets
  • Multivariate forecasting where series share common dynamics
  • Evaluating generalist forecasting vs domain-specific baselines
  • Automated forecasting in data pipelines without per-series tuning
  • Probabilistic prediction with quantile outputs

Pros

  • Multi-frequency support handles hourly through yearly series in one model
  • Patch-based tokenization captures local temporal patterns efficiently
  • Trained on diverse domains, reducing distribution mismatch for unseen series
  • Published with full UNI2TS reproducibility code

Cons

  • CC-BY-NC-4.0 license prohibits commercial use without separate licensing
  • Multivariate support requires careful channel alignment across series
  • Context window bounded; very long historical series require truncation or aggregation
  • Quality on domain-specific series may lag fine-tuned models

Tags

transformerssafetensorstime seriesforecastingpretrained modelsfoundation modelstime series foundation modelstime-seriestime-series-forecastingarxiv:2402.02592license:cc-by-nc-4.0endpoints_compatibleregion:us