tpu24 min read
Optimizing Muse Spark Open on TPU v7 with SGLang-TorchTPU
TL;DR We serve Meta’s Muse Spark Open, an 816B-parameter mixture-of-experts model, on a single TPU v7 host of four chips, with SGLang running on the TorchTPU PyTorch backend. At the... The post Optimizing Muse Spark Open on TPU v7 with SGLang-TorchTPU appeared first on PyTorch.
By Haolin Fu, Xiaoyu Zhang, Junyan Xu, Wangyuan Zhang, Haoguang Cai, RadixArk TeamPyTorch Blog

Read the full story
Originally published by PyTorch Blog on Nov 10, 2026.
Comments
Sign in to join the discussion.