1
Consider the following statements regarding RAG (Retrieval-Augmented Generation) and Fine-Tuning:
- RAG is better suited for scenarios requiring real-time data updates compared to Fine-Tuning.
- Fine-Tuning is more resource-intensive as it requires GPU compute for retraining model weights.
- RAG changes the internal weight parameters of the model to incorporate new information.
Which of the statements given above are correct?
A
1 and 2 only
B
2 and 3 only
C
1 and 3 only
D
1, 2, and 3
Explanation
🌟 Quick Verification
- Direct Answer: The correct answer is 1 and 2 only.
- Core Concept: RAG is a dynamic, low-resource method for grounding models, whereas Fine-Tuning is a static, high-resource method for adjusting model behavior.
🧠 Concept Blueprint
- Statement 1 (Correct): RAG retrieves context from external databases, allowing for real-time updates. Fine-Tuning requires full retraining cycles, making it slow for data updates.
- Statement 2 (Correct): Fine-Tuning involves updating the model's internal parameters, which is computationally expensive and requires significant GPU resources.
- Statement 3 (Incorrect): RAG does not change the model's internal weight parameters. It only appends retrieved context to the prompt. Fine-Tuning is the process that modifies weights.
📊 Fact Spotlight
| Feature | RAG | Fine-Tuning |
|---|---|---|
| Weight Modification | No | Yes |
| Data Freshness | High (Real-time) | Low (Retraining needed) |
| Resource Cost | Low | High |
💡 Exam Strategy & Trivia
Tip
Common Pitfall: Remember that RAG is about retrieval, not retraining. If you see a question about 'updating weights' for RAG, it is incorrect.
Memorable Trivia: RAG is often used in corporate environments where internal documents change daily.