Force CPU-only PyTorch to avoid 8GB CUDA libs on t3.small

Dockerfile

@@ -7,7 +7,13 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 
 WORKDIR /app
 
-# Install Python deps first (cached layer)
+# Install CPU-only PyTorch FIRST to prevent sentence-transformers
+# pulling the massive CUDA-enabled build (~8GB → ~800MB)
+RUN pip install --no-cache-dir \
+    torch==2.3.1+cpu torchvision==0.18.1+cpu \
+    --extra-index-url https://download.pytorch.org/whl/cpu
+
+# Install remaining Python deps
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
@@ -20,7 +26,6 @@ COPY .streamlit/config.toml /app/.streamlit/config.toml
 
 EXPOSE 8501
 
-# Health check
 HEALTHCHECK CMD curl --fail http://localhost:8501/_stcore/health || exit 1
 
 CMD ["streamlit", "run", "streamlit_app.py", \