Instructions to use NX-AI/TiRex-2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- TiRex-2
How to use NX-AI/TiRex-2 with TiRex-2:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
TiRex-2: Generalizing TiRex to Multivariate Data and Streaming
π News (Sep 2026): TiRex-2 has been accepted as an Oral presentation (0.36%) at NeurIPS 2026!
This repository provides the pre-trained multivariate forecasting model TiRex-2 introduced in the paper TiRex-2: Generalizing TiRex to Multivariate Data and Streaming.
TiRex-2 PRO: This repository is our open-source release. TiRex-2 PRO extends it with dedicated support, streaming, hardware-optimized inference (edge, embedded, and industrial PCs, among others), fine-tuning, and classification & regression support β see TiRex-2 PRO below or contact us at contact@nx-ai.com.
TiRex-2
TiRex-2 is a pretrained time series foundation model that forecasts one or many target variates directly from their history, optionally conditioned on past and future-known covariates. A single checkpoint serves both univariate and multivariate forecasting, built on a recurrent architecture designed for efficient streaming settings β all zero-shot, with no task-specific training or fine-tuning.
TiRex-2 generalizes our original univariate model, TiRex, to multivariate forecasting with past and future covariates.
Key facts
Zero-shot multivariate forecasting: TiRex-2 forecasts multiple target variates out of the box, without training or fine-tuning on your data.
Past and future-known covariates: TiRex-2 natively conditions on past covariates and future-known covariates, such as calendar features, holidays, promotions, or scheduled interventions.
Small active footprint: TiRex-2 activates 38.4M parameters in univariate mode and an additional 44.1M parameters for multivariate forecasting.
Documentation
The full documentation, including how-to guides, deployment and the API reference, is available at https://nx-ai.github.io/tirex-2/.
Installation
Requires Python 3.11 or newer. Install PyTorch for your CPU or GPU, then TiRex-2:
pip install tirex-2
Model files
This repository provides the TiRex-2 weights in several formats:
| File | Format | Notes |
|---|---|---|
model.safetensors + config.json |
safetensors | Default. Loaded by load_model("NX-AI/TiRex-2"), including in the Docker images. Avoids unpickling and loads faster. |
model.onnx |
ONNX | Run with only numpy and onnxruntime, no PyTorch needed. See the ONNX guide. |
model.ckpt + model-config.yaml |
PyTorch checkpoint | Legacy format, kept for older releases. Not supported by tirex-2 0.5.0 and later; use tirex-2<0.5.0 if you need it. |
FAQ
See the FAQ in the documentation.
Agent plugin
Forecast time series without training or writing code in Claude Code or Codex using the TiRex-2 plugin:
Claude Code
claude plugin marketplace add NX-AI/tirex-2
claude plugin install tirex2@tirex2-plugins
Codex
codex plugin marketplace add NX-AI/tirex-2
codex plugin add tirex2@tirex2-plugins
Start a new session and ask to forecast with TiRex-2.
Getting started
The most easy way for you to get started is by checking out our "Getting Started" notebook. Moreover, you can jump straight into testing out TiRex-2 using Google Colab. If you have cloned this repository, you can also easily start the notebook via Pixi by running:
pixi run notebook
Minimal usage predicting a simple sine wave
import torch
from tirex2 import TimeseriesType, load_model
from tirex2.plotting import plot_multivariate # requires matplotlib to be installed
# load model
model = load_model("NX-AI/TiRex-2", device="cpu") # use `device="cuda"` if cuda is available
# generate data - target expects time series of shape (n_targets, context_length)
context = torch.sin(torch.arange(128).float() / 8)
ts = TimeseriesType(target=context.unsqueeze(0), past_covariates=None, future_covariates=None)
# perform forecast - each forecast is of shape (n_targets, 9 quantiles, prediction_length)
forecast = model.forecast([ts], prediction_length=32, output_type="numpy")[0]
# visualize result
fig = plot_multivariate(ts, forecast, engine="matplotlib")
fig.show()
Covariate example
This example originates from the "Getting Started" notebook, showing the value of additional covariates.
from tirex2 import load_model
from tirex2.demo import Demo, plot_demo_forecast
# load model
model = load_model("NX-AI/TiRex-2", device="cpu") # use `device="cuda"` if cuda is available
# load data
demo = Demo.create_nonstationary_demo()
ts_univariate = demo.to_timeseries_type(include_covariates=False)
ts_multivariate = demo.to_timeseries_type(include_covariates=True)
# perform forecast - each forecast is of shape (n_targets, 9 quantiles, prediction_length)
forecasts = model.forecast(
timeseries=[ts_univariate, ts_multivariate],
prediction_length=demo.horizon,
output_type="numpy",
)
# visualize result
fig = plot_demo_forecast(demo, *forecasts, engine="matplotlib")
fig.show()
Benchmarking
To reproduce our results for the GIFT-Eval and fev-bench leaderboards, follow the instructions in examples/gifteval and examples/fevbench, respectively.
TiRex-2 Docker image
For detailed instructions on building and running TiRex-2 in a Docker container, see the Docker README.
TiRex-2 PRO
TiRex-2 already provides state-of-the-art performance for zero-shot prediction, so you can use this open-source release without training on your own data.
TiRex-2 PRO extends this open-source release with:
- Support: dedicated support and engagement from NXAI.
- Streaming: incremental forecast updates as new observations arrive, without recomputing forecasts over the full history.
- Speed: optimized inference, including hardware-specific optimizations for edge, embedded, and industrial PC deployments.
- Fine-tuning: models adapted to your data or pretrained on different datasets for improved performance.
- Classification & Regression: TiRex-2 adapted for classification and regression tasks.
See the full feature comparison in the docs, or contact us at contact@nx-ai.com.
Cite
If you use TiRex-2 in your research, please cite our work:
@inproceedings{podest2026tirex2,
title = {{TiRex-2}: Generalizing {TiRex} to Multivariate Data and Streaming},
author = {Podest, Patrick and Pichler, Marco and B{\"u}rger, Elias and Z{\'o}lyomi, Levente and Voggenberger, Bernhard and Berghammer, Wilhelm and Klotz, Daniel and B{\"o}ck, Sebastian and Klambauer, G{\"u}nter and Hochreiter, Sepp},
booktitle = {Advances in Neural Information Processing Systems},
year = {2026},
eprint = {2607.01204},
archivePrefix = {arXiv},
primaryClass = {cs.LG},
url = {https://arxiv.org/abs/2607.01204}
}
Other versions
Alongside this pretrained checkpoint, we release decontaminated versions to enable fair zero-shot evaluation on specific benchmarks by excluding their data from pretraining:
- TiRex-2-g: excludes any overlap with the GiftEval datasets (pretrain and evaluation) from pretraining.
- TiRex-2-gp: includes the GiftEval-Pretrain collection in the pretraining corpus (for comparison against TiRex-2-g).
- TiRex-2-f: excludes all fev-bench eval datasets from pretraining, using the same approach as for GiftEval.
Intended use and limitations
This release supports time-series forecasting as documented. High-risk applications under EU AI Act Article 6 are outside NXAI's intended purpose. NXAI has not validated this release for high-risk compliance. Users and integrators must assess suitability and meet their legal obligations. Unlawful uses, including those prohibited by EU AI Act Article 5, are not permitted. See the Intended Use Notice for details. It does not change the licence or statutory obligations.
License
TiRex-2 is licensed under the Apache License 2.0.
- Downloads last month
- 3,769

