[
{
"query": "What are the main challenges in RAG systems?",
"expected_topics": ["retrieval", "generation", "hallucination", "context"]
},
{
"query": "How does RLHF improve language models?",
"expected_topics": ["reinforcement learning", "human feedback", "alignment"]
},
{
"query": "Explain transformer attention mechanisms",
"expected_topics": ["attention", "self-attention", "transformer", "query", "key", "value"]
},
{
"query": "What are different fine-tuning approaches for LLMs?",
"expected_topics": ["fine-tuning", "LoRA", "adaptation", "training"]
},
{
"query": "How do prompt engineering techniques work?",
"expected_topics": ["prompt", "engineering", "instruction", "few-shot"]
}
]
