Update Dockerfile

Dockerfile

@@ -1,30 +1,32 @@
+# Stage 1: Grab the server binary from the official server image
+FROM ghcr.io/ggml-org/llama.cpp:server AS source
+
+# Stage 2: Setup your environment using the full image (which has the OS tools you need)
 FROM ghcr.io/ggml-org/llama.cpp:full
 
-# 1. Setup Environment
+# 1. COPY the missing binary from Stage 1 to your system path
+# The server image keeps it at /llama-server (root), so we copy it to /usr/local/bin
+COPY --from=source /llama-server /usr/local/bin/llama-server
+
+# 2. Setup Environment
 WORKDIR /app
 ENV HF_HOME=/app/.cache
 ENV NUMBA_CACHE_DIR=/app/.cache
 
-# 2. Install Python & Fix Permissions Tools
+# 3. Install Python & Fix Permissions Tools
 RUN apt-get update && \
     apt-get install -y python3 python3-pip dos2unix && \
     rm -rf /var/lib/apt/lists/*
 
-# 3. FIX: Move the llama-server binary to a global path
-# The 'full' image builds binaries in /llama.cpp/build/bin/
-RUN cp /llama-server /usr/local/bin/llama-server
-
 # 4. Install Python Libraries
 RUN pip3 install --no-cache-dir huggingface_hub
 
 # 5. Copy and Fix the Start Script
 COPY start.sh /app/start.sh
-
-# FIX: Convert Windows line endings to Unix and make executable
 RUN dos2unix /app/start.sh && chmod +x /app/start.sh
 
-# FIX: Give permission to the folder and create cache directory
+# 6. Permissions
 RUN mkdir -p /app/.cache && chmod -R 777 /app
 
-# 6. Start
+# 7. Start
 ENTRYPOINT [ "/app/start.sh" ]