jonpry

It would be nice if OneOf expressions passed some information to the visitor about which rule succeeded. This would be extremely useful in codegen. Perhaps OneOf would create a tuple...

Even though #18 mentions to use fuse values 0x42, 0xD9 and 0xFF. I believe this lfuse value is incorrect as that would set the system up to use the internal...

We have our own compiler system and don't use TensorRT. For individual 2d convolutions, cudnn/cutlass usually give us the best results. Structured sparsity is a fairly easy thing for us...

```# Copyright (c) Tile-AI Corporation. # Licensed under the MIT License. import torch import tilelang from tilelang.autotuner import * import tilelang.language as T import itertools import argparse from tilelang.carver.template import...