up

Dockerfile

@@ -15,12 +15,11 @@ RUN apt-get update && apt-get install -y \
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
+# Use latest llama-cpp-python — 0.3.8 does NOT support Gemma 4 MoE (A4B) architecture
 RUN CMAKE_BUILD_PARALLEL_LEVEL=4 \
     CMAKE_ARGS="-DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS" \
-    pip install --no-cache-dir "llama-cpp-python==0.3.8"
+    pip install --no-cache-dir "llama-cpp-python"
 
-# Model is downloaded at runtime by app.py via hf_hub_download (handles retries/resume)
-# Do NOT wget here — large files fail silently during Docker build on HF Spaces
 RUN mkdir -p /app/models
 
 COPY app.py .