|
|
--- |
|
|
license: mit |
|
|
language: |
|
|
- en |
|
|
base_model: ibm-granite/granite-timeseries-patchtst |
|
|
tags: |
|
|
- time-series |
|
|
- forecasting |
|
|
- patchtst |
|
|
- electricity |
|
|
- eirgrid |
|
|
- energy |
|
|
metrics: |
|
|
- mse |
|
|
- mae |
|
|
library_name: transformers |
|
|
pipeline_tag: time-series-forecasting |
|
|
--- |
|
|
# PatchTST EirGrid Forecaster ⚡ |
|
|
|
|
|
This is a **PatchTST** model fine-tuned on **EirGrid (Irish Grid) System Data** to forecast electricity demand. |
|
|
|
|
|
## Model Details |
|
|
- **Architecture:** PatchTST (Transformer-based Time Series) |
|
|
- **Base Model:** `ibm-granite/granite-timeseries-patchtst` |
|
|
- **Task:** Long-term Forecasting (96-hour horizon) |
|
|
- **Input:** 512 hours of historical load data. |
|
|
- **Output:** 96 hours of future load forecast. |
|
|
|
|
|
## Usage |
|
|
```python |
|
|
from transformers import PatchTSTForPrediction |
|
|
|
|
|
# Load the model |
|
|
model = PatchTSTForPrediction.from_pretrained("Priyansu19/patchtst-eirgrid-forecaster") |
|
|
|
|
|
|
