ParentFull threadqeternity·Yes, absolutely in deep learning. Custom fused CUDA kernels everywhere.View on HN