Alembic Technologies reposted this
“Composite AI, or combining multiple AI techniques, will emerge as the next step in extending the AI revolution. Because Large Language Models (LLMs) are starting to hit natural performance ceilings when used in isolation, technologists will turn to Composite AI architectures that use LLMs as the orchestrating layer, seamlessly integrating them with specialized AI components like knowledge graphs, symbolic reasoning engines, and traditional machine learning models." https://lnkd.in/g6rysXjU
The other term I see a lot is "constellations" of LLMs and mixture of experts that sister into each other and a primary LLM to combine, rank and synthesize the final response. NVIDIA is doing this now through our agent blueprints leveraging NIMs
Thanks for the prediction!! We are seeing this already in 2024 Abby. A clustering of >1 AI Agents each deployed to perform “compliance-first” packaged business capabilities is returning real ROI business value when served via a foundational composable architecture platform, and by this definition would be regarded as a pure play Agentic AI Framework. Would you agree? 🤔 🤖 ⚙️ 🤖 ⚙️ 🤖
Love this
Building foundation models, predictive/time-series, and agent frameworks to help our customers realize the promise of generative AI at NVIDIA
4dNot every node is going to be an LLM in this graph. Classification models using XGBoost, time series transformers for predictive work, small trained models where latency is key but the domain is easy, larger tuned models for super complex stuff where latency isn't as important. Model cost routing when cost optimization becomes a concern at scale. Org structures / multidisciplinary teams... of agents :)