NLP/Prompt & Problem Solving

A Survey of Prompt Engineering Methods in LLMs for Differenct NLP Tasks (2024) 논문 리뷰

아르카눔 2025. 5. 6. 16:05

A Survey of Prompt Engineering Methods in LLMs for Differenct NLP Tasks 논문 (링크)

 

저자는  Shubham Vatsal와 Harsh Dubey 두 사람이다. 

 

서베이 논문인데 여러가지 prompting 방법들의 목록을 간단하게 정리하고 표를 올리고자 한다.

 

Prompt Engineering Techniques

 

  1. Basic / Standard / Vanilla Prompting
  2. Chain-of-Thoguht (CoT)
  3. Self-Consistency
  4. Ensemble Refinement (ER)
  5. Automatic Chain-of-Thoguht (Auto CoT)
  6. Complex CoT
  7. Program-of-Thoughts (PoT)
  8. Least-to-Most
  9. Chain-of-Symbol (CoS)
  10. Structured Chain-of-Thoguht (SCoT)
  11. Plan-and-Solve (PS)
  12. MathPrompter
  13. Contrastive CoT / Contrastive Self-Consistency
  14. Federated Same / Different Parameter Self-Consistency / CoT (FED-SP / DP-SC / COT)
  15. Analogical Reasoning
  16. Synthetic Prompting
  17. Tree-of-Thoguhts (ToT)
  18. Logical Thoughts (LoT)
  19. Maieutic Prompting
  20. Verify-and-Edit (VE)
  21. Reason + Act (ReAct)
  22. Active-Prompt
  23. Thread-of-Thought (ThoT)
  24. Implicit Retrieval Augmented Generation (Implicit RAG)
  25. System 2 Attention (S2A)
  26. Instructured Prompting
  27. Chain-of-Verification (CoVe)
  28. Chain-of-Knowledge (CoK)
  29. Chain-of-Code (CoC)
  30. Program-Aided Language Models (PAL)
  31. Binder
  32. Dater
  33. Chain-of-Table
  34. Decomposed Prompting (DecomP)
  35. Three-Hop Reasoning (THOR)
  36. Metacognitive Prompting (MP)
  37. Chain-of-Event (CoE)

 

Taxonomy Diagram of Prompt Engineering Methods Across Different NLP Tasks