install flash attention

Dockerfile

@@ -4,12 +4,16 @@ FROM ghcr.io/astral-sh/uv:0.2.12 as uv
 # Stage 2: Main application image
 FROM nvidia/cuda:12.1.1-cudnn8-devel-ubuntu22.04
 
+ENV DEBIAN_FRONTEND=noninteractive
+
 # Copy uv from first stage
 COPY --from=uv /uv /uv
 
 # Install Python, pip, venv, and system dependencies
 RUN apt-get update && \
-    apt-get install -y python3.11 python3.11-venv python3-pip ffmpeg && \
+    apt-get upgrade -y && \
+    apt-get install -y --no-install-recommends \
+    python3.11 python3.11-venv python3-pip ffmpeg && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 
@@ -45,9 +49,10 @@ COPY --chown=user requirements.txt .
 RUN --mount=type=cache,target=$UV_CACHE_DIR,uid=1000,gid=1000 \
     /uv pip install -r requirements.txt
 
-# Install flash-attn from pre-built wheel
+# Install flash-attn
 RUN --mount=type=cache,target=$UV_CACHE_DIR,uid=1000,gid=1000 \
-    /uv pip install https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl
+    /uv pip install flash-attn --no-build-isolation
+    #/uv pip install https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl
 
 # Copy application code
 COPY --chown=user . .