CC-Time: Cross-Model and Cross-Modality Learning with LLMs for Time Series Forecasting

Requirements

To install all dependencies:

pip install -r requirements.txt

Datasets

You can access the well pre-processed datasets from Google Drive, then place the downloaded contents under ./dataset.

Quick Demos

1. Download datasets and place them under ./datasets.
2. Create a folder called gpt2 and download the pre-trained parameters of GPT2 from hugging_face.

   mkdir gpt2
   python load_gpt2.py
   

3. Training and testing. Training and testing. Using the ETTm1 dataset as an example, run the following script:

   python main.py