# SysManage translation service (run on the GPU "beast" box).
# The service itself is light — it offloads all model work to Ollama over HTTP,
# so there is NO torch / transformers / CUDA dependency here.  Install Ollama
# separately (see README.md) and pull the translation model.
fastapi>=0.110
uvicorn[standard]>=0.29
httpx>=0.27
pydantic>=2.6
