explicitly pin flash attention 1 to v1.0.9

Author: winglian

Diff for: docker/Dockerfile-base

@@ -38,8 +38,9 @@ WORKDIR /workspace
 
 ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
 
-RUN git clone https://github.com/HazyResearch/flash-attention.git && \
+RUN git clone https://github.com/Dao-AILab/flash-attention.git && \
     cd flash-attention && \
+    git checkout v1.0.9 && \
     python3 setup.py bdist_wheel && \
     cd csrc/fused_dense_lib && \
     python3 setup.py bdist_wheel && \