Google Developers BlogLabs
Integrating TorchTPU into TorchTitan and Optimizing Models for TPU

TorchTPU integrates a native torch.device("tpu") backend directly into PyTorch, allowing developers to train large-scale models on Google Cloud TPUs without rewriting code in JAX.
Read at Google Developers Blog ↗Related

LabsAnnouncing New vLLM and SGLang Inference Engines on TPU Google Developers Blog

LabsML Drift: Next-Gen GPU AI/ML Inference at the Edge Google Developers Blog

LabsAccelerating Spatio-Temporal Attention for Video Diffusion on TPUs Google Developers Blog

LabsReproducing Olmo 3 7B Pre-training in MaxText: case study of large scale training on TPUs Google Developers Blog

Valley"I am giving you money!" Nvidia's CEO wants every country to build its own AI Axios Technology
