# thuml/iTransformer

Official implementation for "iTransformer: Inverted Transformers Are Effective for Time Series Forecasting" (ICLR 2024 Spotlight)

Repository: https://github.com/thuml/iTransformer
Canonical: https://ross.abutalabs.com/products/itransformer
Homepage: https://arxiv.org/abs/2310.06625
Language: Python
License: MIT
License Family: permissive
Topics: time-series-forecasting, transformer
Last push: 2025-07-17T15:53:48+00:00

## Health v2 (maintenance only)
Score: 41/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 32, release rhythm 35, longevity 74
- inputs: {"age_days": 1049, "days_push": 412, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2200, forks 369 (observed 2026-08-28T04:06:25.594110+00:00)

## What it is
Official PyTorch implementation of iTransformer, an ICLR 2024 Spotlight paper that inverts the Transformer architecture for multivariate time series forecasting by treating each variate as a token. It includes experiment scripts and achieves state-of-the-art results on standard forecasting benchmarks.

## Use cases
- forecast multivariate time series with transformers
- reproduce iTransformer paper experiments
- benchmark forecasting models on time series datasets
- predict future values of multiple correlated sensors
- compare transformer baselines for long-term forecasting

## When to choose
- you need a strong transformer-based multivariate forecasting baseline
- you want to reproduce or extend the iTransformer paper
- your data has many correlated variates and long lookback windows

## When to avoid
- you need simple univariate forecasting where linear models suffice
- you need a production-ready forecasting service rather than research code
- you lack GPU resources for training transformer models

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning
- domain: time-series, machine-learning, data-science
- platform: python
- tags: time-series-forecasting, transformer, research-code, iclr-2024, multivariate-forecasting, gpu

## Member repositories
- thuml/iTransformer (main) score 41

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:25.594110+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:47:06.247125+00:00, confidence not recorded.
  - readme: https://github.com/thuml/iTransformer (fetched 2026-08-28T04:06:25.594110+00:00, sha 3665cd07ef37)
  - homepage: https://arxiv.org/abs/2310.06625 (fetched 2026-08-29T10:27:18.809169+00:00, sha 0882d7180727)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T10:27:18.812906+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T10:27:18.816976+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T10:27:18.818832+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T10:27:18.815158+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.