1. Zero-shot learning
- Definition: The model performs a task without seeing any examples of it beforehand.
- Example: You ask a language model to “translate English to French: ‘Hello’” without showing it any examples of translations.
- Key idea: Relies on generalization and pretrained knowledge.
2. One-shot learning
- Definition: The model is given one example of the task before being asked to do a similar one.
- Example: You show the model:
"Translate English to French: 'Good morning' -> 'Bonjour'"
Then ask: "Translate English to French: 'Hello' -> ?"
3. Few-shot learning
- Definition: The model is given a few examples (typically 2-10) to understand the task before trying it.
- Example: Several input-output pairs are shown before the actual query.
4. Fine-tuning / Many-shot learning
- Definition: The model is trained on many labeled examples to learn a task more robustly.
- Used when: You want high accuracy and can afford extra computation and data labeling.
AI Inference
