Skip to content

Commit 3f9d049

Browse files
author
DevakiBolleneni
committed
try build flashattention
1 parent 79da555 commit 3f9d049

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

pytorch/training/docker/2.9/py3/cu130/Dockerfile.arm64.gpu

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -240,7 +240,7 @@ ENV LD_PRELOAD="/usr/local/lib/libnccl.so"
240240
ENV NVTE_FRAMEWORK=pytorch
241241
# Install flash-attn using instructions from https://github.com/Dao-AILab/flash-attention#installation-and-features
242242
# Set MAX_JOBS=4 to avoid OOM issues in installation process
243-
RUN MAX_JOBS=8 pip install --no-cache-dir flash-attn==${FLASH_ATTN_VERSION} --no-build-isolation --verbose
243+
RUN MAX_JOBS=4 pip install --no-cache-dir flash-attn==${FLASH_ATTN_VERSION} --verbose
244244
# Install TE using instructions from https://docs.nvidia.com/deeplearning/transformer-engine/user-guide/installation.html
245245
RUN pip install --no-cache-dir git+https://github.com/NVIDIA/TransformerEngine.git@release_v${TE_VERSION} --no-build-isolation
246246

0 commit comments

Comments
 (0)