Training

2026-10-22 / 22 words

• The Ultra-Scale Playbook: Training LLMs on GPU Clusters
• Granite 4.1 LLMs: How They’re Built
• Notes on Midtraining
• Frontier Pre-training at Microsoft AI