flashinfer.fused_moe.convert_to_block_layout

flashinfer.fused_moe.convert_to_block_layout(input_tensor: Tensor, blockK: int) → Tensor