lighthouse

Linalg/XeGPU Layer norm example (#186)

[KB] bf16 element-wise is done at F32 on CPU

[KB] Linalg generic fusion to help some bad lowering from torch-mlir (#191)

[xegpu][examples] Add truncate-c option to matmul example (#189)

[example] Linalg to XeGPU fused attention implementation. (#153)