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.

Home Page:https://docs.nvidia.com/deeplearning/transformer-engine/user-guide/index.html

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

`inv_freq` of `RotaryPositionEmbedding` is hard-coded to 10k

shijie-wu opened this issue · comments

theta in inv_freq of RotaryPositionEmbedding is hard-coded to 10k

inv_freq = 1.0 / (
10000
** (
torch.arange(0, dim, 2, dtype=torch.float32, device=torch.cuda.current_device())
/ dim
)
)

@sudhakarsingh27 Could you take a look at it?