What transfer learning is

Transfer learning means:

• start from a model pre-trained on a large dataset
• fine-tune it (or use it as a feature extractor) for your task

This is common in:

• computer vision
• NLP

Why it works

Pre-trained models learn general features:

• edges → textures → shapes (vision)
• syntax/semantics patterns (language)

Common workflow

  flowchart TD
  A[Pre-trained model] --> B[Freeze early layers]
  B --> C[Replace final layer]
  C --> D[Train on your dataset]
  D --> E[Optional fine-tune]

When to use it

Use transfer learning when:

• you have limited labeled data
• your domain is similar to the pretraining domain

Mini-checkpoint

What’s the advantage of freezing layers initially?

• prevents destroying learned features and reduces training cost.