3 ms·Running PyTorch Models on Apple Silicon GPUs with the ExecuTorch MLX Delegate2 points by tosh 6mo ago