https://github.com/xuehaouwa/lm-load-forecasting

https://github.com/xuehaouwa/lm-load-forecasting

Last synced: about 1 month ago
JSON representation

• Host: GitHub
• URL: https://github.com/xuehaouwa/lm-load-forecasting
• Owner: xuehaouwa
• Created: 2023-09-19T23:20:52.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-09-20T09:50:59.000Z (over 1 year ago)
• Last Synced: 2023-10-20T09:09:21.688Z (about 1 year ago)
• Size: 1.95 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# LM-Load-Forecasting

## Introduction

This folder includes the code and more details of our paper: Utilizing Language Models for Energy Load Forecasting, which will be presented at BuildSys 2023.

This work is developed from our previous work: [PromptCast: A New Prompt-based Learning Paradigm for Time Series Forecasting](https://arxiv.org/abs/2210.08964)

Its code is also available [here.](https://github.com/HaoUNSW/PISA) 

## How to USE

1. prompt your data (describe your energy usage csv data to language sentences): use `data_prompting.py`

2. prepare train/val/test set data: use `prepare_hf.py`

3. fine-tune language models: use `run_hf_s2s.py`

4. test your fine-tuned language models with your test set: user `run_inference.py`

* For step 3 and 4, there are examples provided in `example.sh`

* Our PromptCast Repo also provides more details, you can check [here.](https://github.com/HaoUNSW/PISA) 

## Note

If you think our paper/code is useful, please cite our paper:

````

@inproceedings{10.1145/3600100.3623730,

author = {Xue, Hao and Salim, Flora D.},

title = {Utilizing Language Models for Energy Load Forecasting},

year = {2023},

publisher = {Association for Computing Machinery},

booktitle = {Proceedings of the 10th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation},

pages = {224–227},

numpages = {4},

location = {Istanbul, Turkey},

series = {BuildSys '23}

}

````

The dataset used in our paper is private data, and we can't share them for now. If we get the green light, I'll modify this repo further