How To Win Friends & Influence LLMs (with Prompt Engineering)
summary
Hands-on demo of Watson X Prompt Lab shows 11 exercises and a working sandbox that connects to models including Granite model, Flan and Llama 70b; free-form chat defaults to 200 tokens (chat ~800), and changing decoding from greedy to sampling plus temperature and repetition-penalty shifts answers from bland to creative. Retrieval Augmented Generation via a vector index ingests PDFs/PowerPoints (limits: PDF 50 MB, PPT 300 MB) to ground responses. Built-in guardrails use a separate microservice classifier for profanity/bias and are enabled by default.
AI-written summary · about
chapters (11)
- 0:00Prompt engineering intro
- 1:26Watson X Prompt Lab setup
- 3:12Tokens and text completion
- 6:48One-shot prompting
- 8:17Model parameters and decoding
- 12:01Choosing models and balancing intelligence and security
- 13:05Retrieval augmented generation
- 16:19Free form vs general chat
- 18:04Chat behavior customization
- 18:26Lab exercises and model options
- 20:34Q&A
related talks
59mTutorial: How 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
46mBeyond Basic Prompting: Supercharging Open Source LLMs with LMQL's Structured Generation
48mBeyond the Prompt: Evaluating, Testing, and Securing LLM Applications
01h 21mCreating Ethical Generative AI for Enterprises
50mBeyond the Prompt: Evaluating, Testing, and Securing LLM Applications