chenyuandGitHub 89117d8b9e use real shift in l2i decomp [pr] (#17080) ...

works for variable shift distace too, also fixed signed arithmetic fill

2026-07-19 13:15:06 -04:00

..

codegen

use real shift in l2i decomp [pr] (#17080)

2026-07-19 13:15:06 -04:00

engine

remove copyin/copyout from Buffer (#17020)

2026-07-14 19:47:22 +03:00

llm

llm: make tokenizer fast (kimi) (#17077)

2026-07-18 17:31:59 -07:00

mixin

more weak dtype materialization raise (#17071)

2026-07-17 23:15:14 -04:00

nn

no weak Tensor from and to real buffer (#17067)

2026-07-17 16:09:10 -04:00

renderer

fix Tensor << Tensor for x86 (#17082)

2026-07-19 12:31:05 -04:00

runtime

llm: make tokenizer fast (kimi) (#17077)

2026-07-18 17:31:59 -07:00

schedule

fix n^2 in limit_bufs by memoizing reachable loads [PR] (#17017)

2026-07-15 23:54:04 -07:00

uop

improve cast to unsign min_max [pr] (#17078)

2026-07-18 21:58:41 -04:00

viz

viz: better sidebar collapse ux (#17072)

2026-07-18 18:07:58 +09:00

__init__.py

start function and add walk rewrite (#14992)

2026-02-25 13:56:27 +08:00

callify.py

expect _offset support, CL and WEBGPU are outliers (#17014)

2026-07-13 18:53:32 -04:00

device.py

remove copyin/copyout from Buffer (#17020)

2026-07-14 19:47:22 +03:00

dtype.py

materialize weak dtype casts to default (#17051)

2026-07-16 16:12:33 -04:00

function.py

Buffer arg -> max_numel [PR] (#16752)

2026-06-25 14:50:20 -04:00

helpers.py

llm: make tokenizer fast (kimi) (#17077)

2026-07-18 17:31:59 -07:00

py.typed

add a single py.typed (#6083)

2024-08-14 17:31:46 -07:00

tensor.py

more weak dtype materialization raise (#17071)

2026-07-17 23:15:14 -04:00