Unraveling AI terminology
summary
Framework separates generative versus predictive and training versus inference, arguing predictive models train in hours–days on company data while generative models need internet‑scale datasets and months of training and GPUs. Claims include 55% developer time saved by code generation and Cloudflare using traffic from ~20% of the web for DDoS prediction. Demo wires LangChain + LangSmith tracing to a RAG pipeline with LLaMA 2 7b on Cloudflare Workers and Cloudflare Vectorize, then swaps in Claude 2 on AWS BedRock to fix routing — it works.
AI-written summary · about
chapters (6)
- 0:00Introduction to AI framework
- 6:18Predictive and generative use
- 11:58Implementation considerations
- 18:15LLM landscape and evaluation
- 26:25Architecting LLM applications
- 34:03Practical demo and conclusions
related talks
53mGenerative AI: Overview and use cases
42mFrom theory to practice: The rise of Generative AI
26mInnovate faster with generative AI
39mUnraveling the Magic of Two Hot Trends: WebAssembly and Generative AI
23mThe State of GenAI & ML in the Cloud Native Ecosystem
27mGenerative AI for technology leaders
26mGenerative AI: Get a Grip … and a Look Under the Hood
49mGenerative AI: Concepts, Tools, and Applications for Data Professionals