Zero-shot Learning
Category: AI Techniques
📖 Definition
Zero-shot learning is an AI's ability to perform tasks it wasn't explicitly trained on. By understanding the task description, the AI can apply its general knowledge to new situations.
🔑 Key Points
- AI can do new tasks without specific training examples
- Relies on general knowledge learned during training
- Prompting with task description enables zero-shot performance
- Demonstrates transfer learning capability
💡 Why It Matters
Zero-shot capability is what makes LLMs so versatile. It means one model can handle thousands of different tasks.