Fix GHCR GPU publish workflow and Docker entrypoint

Author: Gene-Weaver

.github/workflows/publish-gpu.yml

@@ -21,6 +21,28 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v4
 
+      - name: Free disk space (runner)
+        run: |
+          set -eux
+          df -h
+
+          # Remove big preinstalled toolchains from GitHub-hosted runners
+          sudo rm -rf /usr/share/dotnet || true
+          sudo rm -rf /usr/local/lib/android || true
+          sudo rm -rf /opt/ghc || true
+          sudo rm -rf /usr/local/share/boost || true
+          sudo rm -rf /usr/local/share/powershell || true
+          sudo rm -rf /usr/share/swift || true
+
+          # Remove apt lists and any dangling docker data from previous jobs
+          sudo apt-get clean || true
+          sudo rm -rf /var/lib/apt/lists/* || true
+
+          # If docker is already present, prune aggressively
+          docker system prune -af || true
+
+          df -h
+
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v3
 

docker/Dockerfile.gpu

@@ -53,8 +53,15 @@ ENV UV_LINK_MODE=copy
 RUN python3 -m pip install --no-cache-dir -U pip setuptools wheel \
  && python3 -m pip install --no-cache-dir uv
 
-# Install Python deps from requirements
-RUN uv pip install --system -r /app/requirements.txt
+# Install Python deps from requirements (reduce build-time disk usage)
+# - Cache mount prevents cache from being committed into the image layer
+# - Explicit cleanup reduces transient disk pressure and final image size
+ENV UV_CACHE_DIR=/root/.cache/uv
+
+RUN --mount=type=cache,target=/root/.cache/uv \
+    uv pip install --system --no-cache-dir -r /app/requirements.txt && \
+    rm -rf /root/.cache/uv
+
 
 # Required explicit installs (kept per your requirement)
 RUN uv pip install --system \