Deploy FastAPI server with CodeLlama 7B

Dockerfile

@@ -1,20 +1,32 @@
-# Dockerfile for HuggingFace Spaces
-FROM python:3.10-slim
+# Dockerfile for HuggingFace Spaces - Build llama-cpp-python from source
+FROM python:3.11-slim-bookworm
 
 # Set working directory
 WORKDIR /app
 
-# Install system dependencies
+# Install system dependencies including CMAKE build tools
 RUN apt-get update && apt-get install -y \
     build-essential \
-    curl \
+    cmake \
     git \
+    curl \
     && rm -rf /var/lib/apt/lists/*
 
-# Copy requirements and install Python dependencies
+# Copy requirements FIRST (for better caching)
 COPY requirements.txt .
+
+# Install Python dependencies EXCEPT llama-cpp-python
 RUN pip install --no-cache-dir --upgrade pip && \
-    pip install --no-cache-dir -r requirements.txt
+    grep -v "llama-cpp-python" requirements.txt > requirements_temp.txt && \
+    pip install --no-cache-dir -r requirements_temp.txt && \
+    rm requirements_temp.txt
+
+# Build llama-cpp-python from source with CMAKE (this is the key!)
+# Force CMAKE build to ensure it compiles against glibc, not using musl wheels
+RUN CMAKE_ARGS="-DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS" \
+    FORCE_CMAKE=1 \
+    pip install --no-cache-dir --force-reinstall --upgrade --verbose \
+    llama-cpp-python==0.2.90
 
 # Copy application code
 COPY app.py .