Update Dockerfile

Dockerfile

@@ -1,65 +1,64 @@
-FROM python:3.11-slim
 
+# ── Stage 1: copy llama-server out of the official pre-built image ────────────
+FROM ghcr.io/ggml-org/llama.cpp:server AS llamacpp
+ 
+# ── Stage 2: the actual app ───────────────────────────────────────────────────
+FROM python:3.11-slim
+ 
+# libgomp1 is required by llama-server (OpenMP runtime)
 RUN apt-get update && apt-get install -y \
-    git git-lfs ffmpeg libsm6 libxext6 \
-    cmake rsync libgl1 g++ gcc build-essential wget curl \
-    && rm -rf /var/lib/apt/lists/* \
-    && git lfs install
-
+    ffmpeg wget curl libgomp1 \
+    && rm -rf /var/lib/apt/lists/*
+ 
+# Copy the pre-built llama-server binary from stage 1
+COPY --from=llamacpp /llama-server /usr/local/bin/llama-server
+RUN chmod +x /usr/local/bin/llama-server && \
+    llama-server --version 2>&1 | head -1 && \
+    echo "✅ llama-server ready"
+ 
 RUN useradd -m -u 1000 user
-
 WORKDIR /app
-
+ 
 ENV GRADIO_SERVER_NAME="0.0.0.0"
 ENV GRADIO_SERVER_PORT="7860"
 ENV LLAMA_SERVER_URL="http://127.0.0.1:8080"
-
-# ── 1. Download GGUF model ──────────────────────────────────────────────────
+ 
+# ── Download GGUF model at build time ─────────────────────────────────────────
 RUN wget -q --show-progress \
     "https://huggingface.co/unsloth/Qwen2.5-Coder-0.5B-Instruct-128K-GGUF/resolve/main/Qwen2.5-Coder-0.5B-Instruct-F16.gguf" \
     -O /app/model.gguf && \
-    echo "✅ Model size: $(du -sh /app/model.gguf)"
-
-# ── 2. Download pre-built llama-server binary (llama.cpp release b4786) ────
-RUN wget -q --show-progress \
-    "https://github.com/ggerganov/llama.cpp/releases/download/b4786/llama-b4786-bin-ubuntu-x64.zip" \
-    -O /tmp/llama.zip && \
-    unzip /tmp/llama.zip -d /tmp/llama_bin && \
-    find /tmp/llama_bin -name "llama-server" -exec cp {} /usr/local/bin/llama-server \; && \
-    chmod +x /usr/local/bin/llama-server && \
-    rm -rf /tmp/llama.zip /tmp/llama_bin && \
-    echo "✅ llama-server: $(llama-server --version 2>&1 | head -1)"
-
-# ── 3. Python deps ──────────────────────────────────────────────────────────
+    echo "✅ Model: $(du -sh /app/model.gguf)"
+ 
+# ── Python dependencies ───────────────────────────────────────────────────────
 COPY requirements.txt .
 RUN pip install --no-cache-dir --upgrade pip && \
     pip install --no-cache-dir -r requirements.txt
-
-# ── 4. App code ─────────────────────────────────────────────────────────────
+ 
+# ── App code ──────────────────────────────────────────────────────────────────
 COPY --chown=user:user . /app
-
-# ── 5. Startup script: launch llama-server then the Gradio app ─────────────
-RUN echo '#!/bin/bash\n\
-llama-server \
-  --model /app/model.gguf \
-  --host 127.0.0.1 \
-  --port 8080 \
-  --ctx-size 512 \
-  --threads 2 \
-  --n-predict 256 \
-  --no-mmap \
+ 
+# ── Startup script: launch llama-server then Gradio ──────────────────────────
+RUN printf '#!/bin/bash\n\
+set -e\n\
+echo "[startup] Starting llama-server..."\n\
+llama-server \\\n\
+  --model /app/model.gguf \\\n\
+  --host 127.0.0.1 \\\n\
+  --port 8080 \\\n\
+  --ctx-size 512 \\\n\
+  --threads 2 \\\n\
+  --no-mmap \\\n\
   &\n\
-# Wait until the server is ready\n\
-echo "[startup] Waiting for llama-server..."\n\
+echo "[startup] Waiting for llama-server to be ready..."\n\
 for i in $(seq 1 30); do\n\
-  curl -sf http://127.0.0.1:8080/health && break\n\
+  curl -sf http://127.0.0.1:8080/health && echo "" && break || true\n\
   sleep 2\n\
 done\n\
-echo "[startup] llama-server ready, launching app"\n\
+echo "[startup] Launching Gradio app"\n\
 exec python app.py\n\
 ' > /app/start.sh && chmod +x /app/start.sh
-
+ 
 USER user
 EXPOSE 7860
-
+ 
 CMD ["/app/start.sh"]