Hugging Face Releases MoE Training Handbook Built on Megatron-LM

Hugging Face's Nouamane released a Mixture-of-Experts training handbook covering Megatron-LM settings, from hardware checks and dense baselines to MoE layer details and serving/RL, with 55 figures and scripts released. Lessons draw on about 800k GH200 GPU-hours of MoE training on the Alps cluster at CSCS.

• Hugging Face hosts quantization workshop livestream2026-11-05
• OpenAI launches GPT-6 Cyber cybersecurity model2026-11-05
• CCORE Releases GLB-3D-Coder 4B With Full Weights2026-11-05
• Phonon TTS brings high-quality on-device speech, tops benchmarks2026-11-05
• Ripple-65B Preview: U.S. Open-Weight SOTA on Agentic Coding2026-11-04