Chronos Bolt Base Fine Tuned V2

🚀 Model Card for Chronos Bolt BASE Fine-Tuned Model v2

This fine-tuned model is designed for time-series forecasting, especially for intermittent demand forecasts, leveraging the pre - trained capabilities of the base model.

🚀 Quick Start

The model was fine-tuned on a proprietary dataset with 130 million time series datapoints. Although the dataset details are confidential, here are its general characteristics:

• The dataset is composed of multi - dimensional time-series data.
• Additional exogenous information about the target series was added (4 additional columns; 1 of them represents different types of volumes of the target series).

WQL = 0.4530

Summary

The fine-tuned model performs well in intermittent demand forecasts (e.g., Intermittent and Lumpy Time Series demand forecasts for seasonal products).

🔧 Technical Details

Model Architecture and Objective

The model is based on the amazon/chronos-bolt-base architecture and is specifically fine-tuned for intermittent time-series forecasting tasks. It takes advantage of pre-trained capabilities for sequence-to-sequence modeling and is adapted to handle multi-horizon forecasting scenarios.

📚 Documentation

Information Table

📄 Contact

• NIEXCHE
• NIEXCHE (Fevzi KILAS)