Update Dockerfile

Dockerfile

@@ -1,32 +1,21 @@
-# Python 3.10 is required for these specific wheels
-FROM python:3.10-slim
+# Use the ultra-compact pre-compiled image
+FROM samueltallet/alpine-llama-cpp-server:latest
 
-ENV PYTHONUNBUFFERED=1
+# Hugging Face Free Tier settings
+ENV LLAMA_ARG_HOST=0.0.0.0
+ENV LLAMA_ARG_PORT=7860
+ENV LLAMA_ARG_THREADS=2
+ENV LLAMA_ARG_CTX_SIZE=4096
 
-WORKDIR /app
+# Define the Qwen 3.5 model to download and run
+ENV LLAMA_ARG_HF_REPO=HauhauCS/Qwen3.5-4B-Uncensored-HauhauCS-Aggressive
+ENV LLAMA_ARG_HF_FILE=Qwen3.5-4B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf
 
-# Install basic system tools
-RUN apt-get update && apt-get install -y \
-    gcc g++ make cmake git libopenblas-dev wget \
-    && rm -rf /var/lib/apt/lists/*
-
-# 1. Install llama-cpp-python from the official CPU wheel index
-# This skips the 'stuck' compilation and provides the latest architecture support
-RUN pip install --no-cache-dir \
-    --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu \
-    "llama-cpp-python[server]"
-
-# 2. Download the model (Qwen 3.5 4B)
-RUN pip install huggingface_hub
-RUN python3 -c "from huggingface_hub import hf_hub_download; hf_hub_download(repo_id='HauhauCS/Qwen3.5-4B-Uncensored-HauhauCS-Aggressive', filename='Qwen3.5-4B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf', local_dir='.')"
+# Optional: Set an API Key to keep your Space private
+# ENV LLAMA_API_KEY=your_secret_key_here
 
+# Hugging Face needs to know which port to look at
 EXPOSE 7860
 
-# 3. Launch the server
-# Note: n_threads is set to 2 to match the Free Tier CPU limit
-CMD ["python3", "-m", "llama_cpp.server", \
-     "--model", "Qwen3.5-4B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf", \
-     "--host", "0.0.0.0", \
-     "--port", "7860", \
-     "--n_threads", "2", \
-     "--n_ctx", "2048"]
+# The image has its own entrypoint that handles the download and server start
+# We don't need a CMD or ENTRYPOINT here as the base image handles it.