squall/flash-attention
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Bump to v1.0.2

Browse Source
This commit is contained in:
Tri Dao 2023-04-15 22:19:31 -07:00
parent 635f159ee3
commit df1344f866
2 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
setup.py
View File

@ -162,7 +162,7 @@ ext_modules.append(
setup(
name="flash_attn",
version="1.0.1",
version="1.0.2",
packages=find_packages(
exclude=("build", "csrc", "include", "tests", "dist", "docs", "benchmarks", "flash_attn.egg-info",)
),

4
training/Dockerfile
View File

@ -85,11 +85,11 @@ RUN pip install transformers==4.25.1 datasets==2.8.0 pytorch-lightning==1.8.6 tr
RUN pip install git+https://github.com/mlcommons/logging.git@2.1.0
# Install FlashAttention
RUN pip install flash-attn==1.0.1
RUN pip install flash-attn==1.0.2
# Install CUDA extensions for cross-entropy, fused dense, layer norm
RUN git clone https://github.com/HazyResearch/flash-attention \
&& cd flash-attention && git checkout v1.0.1 \
&& cd flash-attention && git checkout v1.0.2 \
&& cd csrc/fused_softmax && pip install . && cd ../../ \
&& cd csrc/rotary && pip install . && cd ../../ \
&& cd csrc/xentropy && pip install . && cd ../../ \