We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
8fea425
CUDA has been enabled.
Gradient evals are now faster, for both OpenCL and CUDA.