TensorRT

NVIDIA Open-Sources AITune: Auto-Selects Fastest PyTorch Inference Backend

NVIDIA releases AITune, an open-source toolkit that benchmarks and auto-picks the optimal inference backend for PyTorch models.