Creating Your Own Downstream Application

Notebooks:

• PyTorch dataset template
  • Download scalers
  • Load configuration
  • Define DS dataset
  • Initialize class without Surya stacks
  • Test length and structure
  • Define dataloader
  • Initialize class with Surya stacks
  • Plotting input stack
  • Define dataloader
  • Conclusions
• PyTorch Lightning baseline template
  • Set your cuda visible device
  • Download scalers
  • Load configuration
  • Define Downstream (DS) datasets
  • Define simple baseline model
  • Define your metrics
  • Define your PyTorch ligthning module
  • Set your global seeds
  • Intialize Lightning module
  • Logging
  • Initialize trainer
  • Fit the model
  • Conclusion
• PyTorch Lightning fine-tuning template
  • Set your cuda visible device
  • Download scalers and Weights
  • Load configuration
  • Define Downstream (DS) datasets
  • Initialize the HelioSpectformer model
  • Load model weights
  • To LoRA or not to Lora
  • Define your metrics
  • Define your PyTorch ligthning module
  • Set your global seeds
  • Intialize Lightning module
  • Logging
  • Initialize trainer
  • Fit the model
  • Conclusion
• PyTorch Lightning fine-tuning template (script)