Accueil Explorer Aide
S'inscrire Connexion
david
/
flash-attention
miroir de https://github.com/Dao-AILab/flash-attention
1
0
Fork 0
Fichiers Tickets 0 Wiki
Aborescence: fb9c9cbbe9
Branches Tags
flash-attention
/
csrc
/
xentropy
Chirag Jain 50896ec574 Make nvcc threads configurable via environment variable (#885) il y a 9 mois
..
README.md 5400fdc4ac [CE] Implement CrossEntropyLoss in Triton il y a 1 an
interface.cpp dff68c2b22 Add smoothing for CrossEntropyParallel, rename to CrossEntropyLoss il y a 2 ans
setup.py 50896ec574 Make nvcc threads configurable via environment variable (#885) il y a 9 mois
xentropy_kernel.cu dff68c2b22 Add smoothing for CrossEntropyParallel, rename to CrossEntropyLoss il y a 2 ans

README.md

This CUDA extension implements optimized cross-entropy loss, adapted from Apex's Xentropy. We make it work for bfloat16 and support in-place backward to save memory.

It has only been tested on A100s.

cd csrc/xentropy && pip install .

As of 2023-09-15, this extension is no longer used in the FlashAttention repo. We've instead switched to a Triton-based implementation. See the CrossEntropyLoss module for more details.