Upload Dockerfile

Dockerfile

@@ -3,9 +3,10 @@ FROM python:3.12-slim
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-# Runtime deps for OpenBLAS (no full build toolchain to speed up)
+# Build & runtime deps for llama-cpp-python (CPU, OpenBLAS)
 RUN apt-get update && apt-get install -y --no-install-recommends \
-    libopenblas-dev curl ca-certificates && \
+    build-essential cmake git curl ca-certificates \
+    libopenblas-dev && \
     rm -rf /var/lib/apt/lists/*
 
 # Workdir
@@ -15,8 +16,8 @@ WORKDIR /app
 COPY requirements.txt /app/requirements.txt
 RUN pip install --upgrade pip && \
     pip install --no-cache-dir -r requirements.txt && \
-    pip install --no-cache-dir --only-binary=:all: llama-cpp-python==0.2.90 || \
-    pip install --no-cache-dir --prefer-binary llama-cpp-python==0.2.90
+    CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" \
+    pip install --no-cache-dir llama-cpp-python==0.2.90
 
 # App
 COPY app.py /app/app.py