Update Dockerfile

Dockerfile

@@ -1,31 +1,35 @@
+# FINAL DOCKERFILE
+
 # Use the standard Python 3.11 image for maximum compatibility
 FROM python:3.11
 
-# Install system dependencies (curl is no longer strictly needed but is useful to have)
-RUN apt-get update && apt-get install -y curl && rm -rf /var/lib/apt/lists/*
+# Install system dependencies needed for compilation
+RUN apt-get update && apt-get install -y \
+    curl build-essential cmake \
+    && rm -rf /var/lib/apt/lists/*
 
 WORKDIR /app
 
 # Create writable directories
 RUN mkdir -p /app/.cache /app/vector_database && chmod -R 777 /app
 
-# Set environment variables
+# --- THE KEY FIX ---
+# Set CMAKE_ARGS to tell llama-cpp-python to build without GPU support,
+# which makes the compilation faster.
 ENV TRANSFORMERS_CACHE=/app/.cache \
     HF_HOME=/app/.cache \
-    CHROMADB_DISABLE_TELEMETRY=true
-
-# Install a specific, VERIFIED pre-built wheel for llama-cpp-python
-RUN pip install --no-cache-dir \
-    https://github.com/abetlen/llama-cpp-python/releases/download/v0.2.75/llama_cpp_python-0.2.75-cp311-cp311-manylinux_2_28_x86_64.whl
+    CHROMADB_DISABLE_TELEMETRY=true \
+    CMAKE_ARGS="-DLLAMA_CUBLAS=OFF"
 
-# Now, install the rest of the dependencies
+# Copy and install Python requirements.
+# THIS STEP WILL BE VERY SLOW. Please allow up to 30 minutes.
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
 # Copy the application code
 COPY . .
 
-# --- REMOVED: The model will now be downloaded by the Python app at runtime ---
+# --- NOTE: The model is downloaded by app.py at runtime ---
 
 # Expose the application port
 EXPOSE 7860