Skip to content

peidaqi/bert_gpu

Repository files navigation

BERT-GPU

A modified version of the Google BERT Tensorflow model for multi-GPU support.

You can simply use the FLAGS.use_tpu to turn GPU support on/off, either in code or command line.

For original Release Note, please refer to the Google BERT repo.

This code uses Tensorflow distribute library and thus requires Nvidia NCCL to run.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages