NVIDIA / TransformerEngine

A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit floating point (FP8) precision on Hopper and Ada GPUs, to provide better performance with lower memory utilization in both training and inference.

Apache License 2.0

1.6k stars 255 forks source link

Description

Expose rotary_base as an arg instead of hardcoding to 10000

Fixes # (issue) https://github.com/NVIDIA/TransformerEngine/issues/849

Type of change

[ ] Documentation change (change only to the documentation, either a fix or a new content)
[ ] Bug fix (non-breaking change which fixes an issue)
[ ] New feature (non-breaking change which adds functionality)
[ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
[ ] Infra/Build change
[x] Code refractor

NVIDIA / TransformerEngine

Expose `rotary_base` as an arg instead of hardcoding #944

Description

Type of change