Tutorial: How to Win Friends & Influence LLMs (with Prompt Engineering)
summary
Hands-on prompt engineering using Watson X and its Granite models shows concrete levers: set min/max tokens, stop sequences, temperature, top-p/top-k and repetition penalty; examples and few-shot prompts dramatically change outputs; use Watson X to load a vector index and ground answers with retrieval-augmented generation (RAG) (demo loads GitHub issue 36122); smaller models perform faster and cheaper — 13B vs 70B — 1300 tokens: $0.80 vs $3; 11 lab exercises teach exact prompt patterns that reproduce those effects.
AI-written summary · about
chapters (16)
- 0:00Workshop introduction
- 1:04Speaker and IBM Watson
- 5:22Watson AI sandbox setup
- 7:07Watson X chat interface
- 10:01Prompt engineering basics
- 14:16Output structure control
- 20:12Parameters and temperature
- 23:41Repetition and model selection
- 27:22Retrieval augmented generation
- 35:15RAG prioritization and guardrails
- 41:49Instruct vs chat models
- 45:44Workshop exercises
- 47:11IBM AI examples
- 49:03Model demo and cost analysis
- 50:33SQL code generation
- 52:19Q&A on data and models
related talks
26mHow To Win Friends & Influence LLMs (with Prompt Engineering)
37mWorkshop: Prompt Engineering with Watsonx
01h 39mMastering LLMs: From Prompt Engineering to Agentic AI
26mMastering Large Language Models: Prompt Engineering for Precision & Performance
19mUnlocking the Full Potential of LLMs: Advanced Prompt Engineering Techniques
13mPrompt Engineering Simplified
30mIs Prompt Engineering Dead? How Auto-Optimization is Changing the Game
38mPrompt Engineering ⚙️ - Addressing the sensitivity of Large language models