Project Description

We introduce LLM-TS Integrator, a framework that effectively integrates the capabilities of LLMs with traditional TS modeling.

Usage

1. Set Up Environment: Ensure Python 3.8 is installed and then set up the required libraries using:

   pip install -r requirements.txt
   

2. Data Preparation: Download pre-processed datasets from Google Drive or Baidu Drive. Place the files in ./dataset. Below is an overview of the supported datasets:

   

3. Model Training and Evaluation: Execute scripts for various tasks using commands in ./scripts/:

   • Long-term forecast: bash ./scripts/long_term_forecast/EXP1.sh
   • Short-term forecast: bash ./scripts/short_term_forecast/EXP1.sh
   • Imputation: bash ./scripts/imputation/ETT_script/EXP1.sh
   • Anomaly detection: bash ./scripts/anomaly_detection/PSM/EXP1.sh
   • Classification: bash ./scripts/classification/EXP1.sh

Acknowledgements

Special thanks to the TimesNet library (TSlib) for their extensive resources that supported this project.