license: cc-by-nc-nd-4.0 base_model: - Salesforce/moirai-1.0-R-small - Salesforce/moirai-1.0-R-base - Salesforce/moirai-1.0-R-large pipeline_tag: time-series-forecasting

Time Series Forecasting Model

This model is built on top of the Salesforce Moirai architecture, specifically fine-tuned and trained from scratch for time series forecasting tasks.

Model Details

Model Description

This model leverages the Moirai architecture from Salesforce, which is designed specifically for time series forecasting. The model is available in multiple versions:

Fine-tuned variants (located in ./finetune directory)
Trained from scratch variants (located in ./train_from_scratch directory)

All checkpoint files (.ckpt) include self-explanatory naming conventions that indicate their configuration and training approach.

Model Type

Time Series Forecasting

Version

1.0

Training Procedure

Training Methodology

Two distinct training approaches were utilized:

Fine-tuning approach: Starting with the pre-trained Moirai models (small, base) and fine-tuning on domain-specific data
Training from scratch: Building models with the Moirai architecture but training entirely on specific datasets

Limitations and Biases

The model inherits any limitations present in the base Moirai architecture
Performance may degrade for extremely long-range forecasts
The model may not perform optimally on domains significantly different from its training data

Additional Information

Checkpoints

The model checkpoints are organized as follows:

Fine-tuned models: ./finetune/*.ckpt
Models trained from scratch: ./train_from_scratch/*.ckpt

Citation

If you use this model in your research, please cite: TBA

License

This model is licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).

bkoyuncu
/

MacroEconModels