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.
https://docs.nvidia.com/deeplearning/transformer-engine/user-guide/index.html
Apache License 2.0
1.6k stars 255 forks source link

[JAX] Add experimental internal used THD(packed) fused attn API #964

Closed zlsh80826 closed 7 hours ago

zlsh80826 commented 1 week ago

Description

This PR exposes underly THD format API to JAX side and wrap it to be supported on multi-gpus.

Fixes # (issue)

Type of change

Changes

Checklist:

zlsh80826 commented 1 week ago

/te-ci jax

zlsh80826 commented 1 week ago

/te-ci jax