Keynote: How LLMs Have Changed the Way Organizations Do AI
summary
Describes concrete lessons from deploying LLMs with Ray and growing hardware fleets, including organizations using dozens, hundreds, or thousands of GPUs; lays out two adoption phases—fast build-and-learn prototyping (focus on iteration, evaluation speed, hybrid API vs fine-tune choices) and a scale phase (customization, pretraining, model routing, latency, reliability, cost); claims that fine-tuned Llama models can beat GPT‑4 at much lower cost and emphasizes tooling gaps for multimodal data and heterogeneous accelerators.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:40Anyscale
- 1:30Challenges
- 6:00Phases
- 9:07Dimensions
- 11:00Fine Tuning
- 11:27Massive Expansion
related talks
12mKeynote: How LLMs Have Changed the Way Organizations Do AI
05mKeynote: Teaching LLMs New Knowledge
46mPlaying the long game
33mOpening Keynote: LLMs, between fantasy and reality - the poin...
28mAnyscale's Unified Platform for LLM Development and Deployment
01h 00mPANEL - Data, Code, and Life with LLMs.
30mLLMs: Beyond the Hype - A Practical Journey to Scale
33mAccelerating the LLM Lifecycle on the Cloud