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).
