PyTorch BlogTools
Training a 1T-Parameter MoE on EU public infrastructure with PyTorch and TorchTitan

TL;DR Scaling pretraining for a ~1-trillion-parameter Mixture-of-Experts (MoE) model on shared, public infrastructure was previously considered unfeasible due to unique architectural and interconnect…
Read at PyTorch Blog ↗More from PyTorch Blog on Accept All

ToolsExecuTorch WebGPU: PyTorch Models on the GPU, in the Browser PyTorch Blog

ToolsOptimizing Muse Spark Open on TPU v7 with SGLang-TorchTPU PyTorch Blog

ToolsCall for Proposals: PyTorch Day China 2026 PyTorch Blog

ToolsKVCache Meetup Korea 2026: Event Recap PyTorch Blog

ToolsAnnouncing the Special Interest Group (SIG) for Ray on Accelerators PyTorch Blog
