Bumblebee + EXLA backend (CPU by default, CUDA/ROCm when configured).
Holds the logic previously inline in ModelManager: model loading through
BumblebeeLoader and inference through the stored serving.
Bumblebee + EXLA backend (CPU by default, CUDA/ROCm when configured).
Holds the logic previously inline in ModelManager: model loading through
BumblebeeLoader and inference through the stored serving.