AI BriefingHow to Build a Frontier Model Factory (48 minutes)
Key point
Poolside's Co-CEO Eiso Kant explains the training infrastructure and strategy behind building Laguna S, a 118B MoE model.
Details
Poolside's Co-CEO Eiso Kant reveals the process behind how a small research team built Laguna S, a model with performance that overwhelms much larger models.
Laguna S is a model with a 118B Mixture-of-Experts (MoE) architecture, and it recorded performance surpassing much larger open-weight systems. The key to this process lies in building efficient Training Infrastructure.
Key points are as follows:
- How to design optimal infrastructure for a small team to train large-scale models
- Architectural decisions for the Laguna S model
- Efficient resource management strategies to achieve large-scale model performance
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Step 5 Preview Released: 600B Parameter MoE Model Highlights Agentic Tasks and Financial Analysis PerformanceDeepSeek Releases DSec: Elastic Sandbox Infrastructure for Large-Scale Agentic Training