The State of the Llama Ecosystem
summary
Llama sees 350M+ downloads and a jump in usage and releases culminating in the 405B model with 128k context and a changed license that explicitly allows training on model outputs; the 405B is used as a teacher to post-train 8B and 70B models and improve them, trained on ~15 trillion tokens across H100s. The ecosystem ships a Llama Stack CLI and repos for quantization, safety (Purple Llama), RAG and agent orchestration.
AI-written summary · about
chapters (6)
- 0:00Introduction and growth
- 1:39Llama 3.1 architecture
- 7:17Model performance evaluation
- 8:41Agentic development tools
- 12:48Llama Stack API and CLI
- 15:35Ecosystem and deployment
related talks
38mMeta's Roadmap for Full Stack AI
12mKeynote: The Llama Ecosystem: Past, Present and Future
48mBringing Llama 3 to Life
31mBuilding LLaMA: Meta's Director of GenAI
16mLlama 4
12mUnpack the Latest Llama 4 Herd
11mIntroducing Llama API: A Seamless Way to Revolutionize Your Workflow
22mThe Disruptive Potential of On-Device LLMs