Is Prompt Engineering Dead?
summary
Automatic prompt optimization claims to replace manual tuning by treating prompts as trainable parameters and ships working frameworks: DSPy, AdelFlow, and TextGrad. Methods shown include bootstrap few‑shot (plus random search), MIPRO with Bayesian optimization, and textual gradients that produce semantic feedback; TextGrad benchmarks match or beat bootstrap few‑shot with eight demonstrations. Demos include pipeline-aware optimization, no‑code MLOps LM ops with an Apache‑2 open beta, and caveats about dependence on meta‑prompts and teacher LLM limits.
AI-written summary · about
chapters (18)
- 0:00Welcome
- 1:51Automatic prompt optimization
- 3:38Single vs compound systems
- 7:11Few-shot learning
- 9:18Choosing evaluation metric
- 10:20Bootstrap few-shot
- 11:49Joint optimization
- 13:16Limitations of feedback
- 13:57Textual gradients
- 15:25Practical frameworks
- 15:45DSPy framework
- 19:00DSPy modules
- 20:27Optimizing with teleprompters
- 21:13TextGrad framework
- 21:34TextGrad concepts
- 22:40Adelflow framework
- 24:50Manual prompt engineering dead?
- 26:41Q&A
related talks
30mIs Prompt Engineering Dead? How Auto-Optimization is Changing the Game
29mIs Prompt Engineering Dead? How Auto-Optimization is Changing the Game
02h 01mPrompt Engineering and AI Red Teaming
31mDSPy: The End of Prompt Engineering as We Know It
13mPrompt Engineering Simplified
39mOptimizing a Prompt for Production
14mPrompt Engineering is Dead
26mMastering Large Language Models: Prompt Engineering for Precision & Performance