ubc-bsd10-track-b

Supplemental materials for UBC BSD10 Track B

module 6 labs

I. Getting setup with api access (15 minutes)

• https://platform.openai.com/settings/organization/api-keys
• testing with curl
• setting up Postman
  • base url
  • required headers
  • body
  • handling sensitive data
  • trouble shooting

II. Understanding LLM out-of-the-box capabilities Prompt: "Is Lawrence McDaniel a good photographer?" (15 minutes)

• Google
• ChatGPT
• Postman via OpenAI/GoogleAI/MetaAI
• curl via OpenAI
  • what's the difference?
  • streaming and other UX improvements
  • tool calling
  • ChatGPT's back door web search

III. Getting setup with Python OpenAI PyPI (10 minutes)

• https://github.com/FullStackWithLawrence/openai-hello-world/
• clone
• make init
• make docker-build
• live demo

module 7 labs

I. LLM Tool Call live example Prompt: "What's the weather in Vancouver right now?" (15 minutes) docs: https://platform.openai.com/docs/guides/function-calling?api-mode=responses example: https://platform.smarter.sh/docs/developer/weather-function/

• break down a tool call in smarter
  • ChatGPT
  • Postman
  • https://openai.lawrencemcdaniel.com/ -> https://github.com/FullStackWithLawrence/aws-openai
  • https://platform.smarter.sh/workbench/example/chat/

II. LLM RAG live example Prompt: "What analytics and accounting courses does Wharton offer?" (30 minutes)

• Google

  • ChatGPT

  • LLM Api prompt via Postman

  • RAG Prompt: https://github.com/FullStackWithLawrence/openai-embeddings

  • setup the project

  • build the vector database

  • demo template technique

  • demo with / without RAG approach

  • show a sample human language query to vector database