[AI Boostcamp Day 13] Saving, Transfering and Monitoring Models

What I studied today:

1. Recent Tendency of Using Models
   • Using pretrained models to fine-tune on a specific task
2. Saving a Model
   • Can either save a the whole model or save only the parameters (smaller model)
   • Practice with Code
3. Saving a Checkpoint
   • Utilizing the function for saving models, we can create a checkpoint at every iterations
4. Transfer Learning
   • Using a pretrained model on a larger dataset and fine-tuning on a specific, smaller dataset to obtain better results
   • Freezing: frozing some of the parameters when tuning a pretrained model
5. Monitoring Tools for PyTorch
   • Tensorboard
   • Weight and Biases