# of sequential models (e.g., Transformers) or neural networks with small batch size. # It takes a vector :math:`x` as input and produces a vector :math:`y` of the same shape as output. # The ...
than PyTorch's native op for a particular class of matrices: those whose rows can fit in the GPU's SRAM. In doing so, you will learn about: * The benefits of kernel fusion for bandwidth-bound ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results