llmnlpproduct
Prompt engineering
System prompts, few-shot, chain-of-thought, structured output — making LLMs produce what you actually want.
Уровни глубины
L0Intro~1ч
Writes a task instruction and zero-shot asks an LLM.
L1Basics~5ч
Uses few-shot examples, system prompts, role instructions; output structured JSON.
L2Working~12ч
Applies CoT, self-consistency, ReAct, Reflexion; caches prompts for production.
L3Advanced~20ч
Evaluates prompts with golden sets; auto-prompts (DSPy, APE); prompt caching.
L4Research~40ч
Contributes to prompt-optimisation theory or emergent reasoning research.
Ресурсы
L1 — Basics