laguna-m1-xs2-technical-report.pdf
poolside.ai · 7,674 words · saved by 2 readers
N/A
Poolside Team Laguna M.1/XS.2 Technical Report We present LAGUNA M.1 and LAGUNA XS.2, two Mixture-of-Experts foundation mod- els built for long-horizon, agentic coding: M.1 has 225.8B total parameters (23.4B activated per token) and XS.2 has 33.4B total (3B activated). Both models were trained from scratch end-to-end inside the same internal system that we refer to as our Model Factory: a tightly-integrated stack of versioned data, training, evalu- ation, and inference components that turn model development into an industrial process. We describe the principles and design choices of the…
saved by
related reading
- Composer2.pdfcursor.com
- frontier model training methodologies | Alex Wa's Blogdjdumpling.github.io
- Inkling: Our Open-Weights Model - Thinking Machines Labthinkingmachines.ai
- GLM-5.2: Built for Long-Horizon Tasksz.ai
- Quantifying infrastructure noise in agentic coding evals \ Anthropicanthropic.com
- MAI-Thinking-1: Building a Hill-Climbing Machinemicrosoft.ai
- MAI-Thinking-1: Building a Hill-Climbing Machinemicrosoft.ai
- Trending Papers - Hugging Facepaperswithcode.com
- The hidden engineering behind foundation model buildingpoolside.ai
- PostTrainBenchposttrainbench.com
- AINews | AINewsnews.smol.ai
- RL Post-Training on Macs | Pluralis Researchpluralis.ai