Hugging Face
Hugging Face’s MoE Handbook Turns Training Experience into Reusable Knowledge
Hugging Face’s MoE handbook distills practices from 800,000 GH200 GPU-hours on Megatron-LM, routing, balancing, deployment, and RL.
Back to home
Megatron-LM
3 articles tagged with this topic
Miles
Chinese Team Open-Sources Miles RL Framework — LLM 'Taming' Tools Go Mainstream
Miles v0.1 open-sources RL post-training engineering. Chinese devs join infrastructure race — 'taming' LLMs no longer just big tech patent.
Megatron-LM
100B-AI Training Pipeline Goes Open Source — Under 10 Chinese Teams Can Run It
Engineers stitched K8s, Megatron-LM, and vLLM into a 100B-parameter training pipeline. Side-by-side docs help—but under 10 Chinese teams can run it.