Notion

What does RAG primarily help with in LLM based applications?

What distinguishes an AI agent from a plain LLM call?

Describe the typical agent execution loop.

When is a fixed LCEL chain preferable to an agent loop, and what is the decisive question to ask?

What LangGraph feature allows the agent loop to cycle back to an earlier node rather than terminating?

Which practices reflect the principle of least privilege applied to an agent's tool set?

List SWE-bench, AgentBench, GAIA, and TAU-bench: what they measure, their shared blind spots, and how to interpret agent benchmark numbers.

Define reward hacking in the RLHF context

How prompt caching pays off disproportionately well in multi-agent systems

What are BPE merge residues (LiteToken) and what do they imply for vocabulary efficiency?