Hugging Face adds native TPU support to Transformers, Diffusers and Kernels
Overview
Hugging Face says Transformers, Diffusers and Kernels now run natively on TPUs through TorchTPU, a PyTorch backend that treats TPU as another device.
The company's blog post shows device_map="tpu" for Transformers and pipe.to("tpu") for Diffusers. Hugging Face says Optimum TPU is no longer required to run these models on TPU.
The post reports that 96.5% of the TPU tests the company runs pass, with 34 reported issues and 19 already fixed. These figures come from Hugging Face's own testing.
Written by AI from the articles below · updated Oct 20, 11:52 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Hugging Face BlogOfficialPickHugging Face explains running Transformers and Diffusers natively on TPUs with TorchTPU
AIHugging Face says TorchTPU adds TPU as a native PyTorch backend, so models can be sent to a TPU device without the older torch_xla path. The post shows device_map="tpu" for Transformers and pipe.to("tpu") for Diffusers, and reports that 96.5% of the TPU tests it runs pass, with 34 reported issues and 19 already fixed.
Heat Trend
- Current comparable heat
- 6
- Peak comparable heat
- 6Oct 20, 12:00 PM ET
- Change in past 24 hours
- –
The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.