Chronos Bolt Base

🚀 Chronos-Bolt⚡ (Base)

Chronos-Bolt is a family of pretrained time series forecasting models designed for zero-shot forecasting. It's based on the T5 encoder-decoder architecture and trained on nearly 100 billion time series observations. It divides historical time series context into patches of multiple observations and inputs them into the encoder. The decoder then directly generates quantile forecasts for multiple future steps, a method called direct multi-step forecasting. Compared to the original Chronos models of the same size, Chronos-Bolt models are more accurate, up to 250 times faster, and 20 times more memory-efficient.

🚀 Quick Start

🚀 Update Feb 14, 2025: Chronos-Bolt models are now available on Amazon SageMaker JumpStart! Check out the tutorial notebook to learn how to deploy Chronos endpoints for production use in a few lines of code.

✨ Features

Performance

The following plot compares the inference time of Chronos-Bolt against the original Chronos models for forecasting 1024 time series with a context length of 512 observations and a prediction horizon of 64 steps.

Chronos-Bolt models are not only significantly faster but also more accurate than the original Chronos models. The following plot reports the probabilistic and point forecasting performance of Chronos-Bolt in terms of the Weighted Quantile Loss (WQL) and the Mean Absolute Scaled Error (MASE), respectively, aggregated over 27 datasets (see the Chronos paper for details on this benchmark). Remarkably, despite having no prior exposure to these datasets during training, the zero-shot Chronos-Bolt models outperform commonly used statistical models and deep learning models that have been trained on these datasets (highlighted by *). Furthermore, they also perform better than other FMs, denoted by a +, which indicates that these models were pretrained on certain datasets in our benchmark and are not entirely zero-shot. Notably, Chronos-Bolt (Base) also surpasses the original Chronos (Large) model in terms of the forecasting accuracy while being over 600 times faster.

Chronos-Bolt models are available in the following sizes.

📦 Installation

Zero-shot inference with Chronos-Bolt in AutoGluon

Install the required dependencies.

pip install autogluon

Deploying a Chronos-Bolt endpoint to SageMaker

First, update the SageMaker SDK to make sure that all the latest models are available.

pip install -U sagemaker

💻 Usage Examples

Basic Usage

Zero-shot inference with Chronos-Bolt in AutoGluon

from autogluon.timeseries import TimeSeriesPredictor, TimeSeriesDataFrame

df = TimeSeriesDataFrame("https://autogluon.s3.amazonaws.com/datasets/timeseries/m4_hourly/train.csv")

predictor = TimeSeriesPredictor(prediction_length=48).fit(
    df,
    hyperparameters={
        "Chronos": {"model_path": "autogluon/chronos-bolt-base"},
    },
)

predictions = predictor.predict(df)

Deploying a Chronos-Bolt endpoint to SageMaker

from sagemaker.jumpstart.model import JumpStartModel

model = JumpStartModel(
    model_id="autogluon-forecasting-chronos-bolt-base",
    instance_type="ml.c5.2xlarge",
)
predictor = model.deploy()

Advanced Usage

Deploying a Chronos-Bolt endpoint to SageMaker and making predictions

import pandas as pd
df = pd.read_csv("https://raw.githubusercontent.com/AileenNielsen/TimeSeriesAnalysisWithPython/master/data/AirPassengers.csv")

payload = {
    "inputs": [
        {"target": df["#Passengers"].tolist()}
    ],
    "parameters": {
        "prediction_length": 12,
    }
}
forecast = predictor.predict(payload)["predictions"]

💡 Usage Tip

For more advanced features such as fine-tuning and forecasting with covariates, check out this tutorial. Chronos-Bolt models can be deployed to both CPU and GPU instances. These models also support forecasting with covariates. For more details about the endpoint API, check out the example notebook.

📚 Documentation

Citation

If you find Chronos or Chronos-Bolt models useful for your research, please consider citing the associated paper:

@article{ansari2024chronos,
    title={Chronos: Learning the Language of Time Series},
    author={Ansari, Abdul Fatir and Stella, Lorenzo and Turkmen, Caner and Zhang, Xiyuan, and Mercado, Pedro and Shen, Huibin and Shchur, Oleksandr and Rangapuram, Syama Syndar and Pineda Arango, Sebastian and Kapoor, Shubham and Zschiegner, Jasper and Maddix, Danielle C. and Mahoney, Michael W. and Torkkola, Kari and Gordon Wilson, Andrew and Bohlke-Schneider, Michael and Wang, Yuyang},
    journal={Transactions on Machine Learning Research},
    issn={2835-8856},
    year={2024},
    url={https://openreview.net/forum?id=gerNCVqqtR}
}

📄 License

This project is licensed under the Apache-2.0 License.