chenyuandGitHub 60215deb60 tiebreak in fold_divmod_congruence (#15190) ...

need to try both direction

2026-03-09 03:40:39 -04:00

..

apps

llm: fix chunked prefill (#15182)

2026-03-07 22:08:31 +08:00

codegen

even more divmod recombine (#15163)

2026-03-08 23:52:26 -04:00

engine

memplanner opt copy bufs (#15110)

2026-03-08 22:28:01 +03:00

mixin

move more to mixins (#14780)

2026-02-16 11:35:00 +08:00

nn

don't use intermediate dict in onnx parse (#15181)

2026-03-07 00:08:03 -05:00

renderer

fix QCOMCLRenderer pickle (#15189)

2026-03-09 00:36:16 -04:00

runtime

QCOM CL compiler prints LLVMIR when DEBUG>=8 (#15187)

2026-03-09 00:15:20 -04:00

schedule

fix symbolic shape call + chunked prefill (#15149)

2026-03-05 14:02:26 +08:00

uop

tiebreak in fold_divmod_congruence (#15190)

2026-03-09 03:40:39 -04:00

viz

viz: exclude redundant traceback from SDMA (#15185)

2026-03-09 05:12:14 +09:00

__init__.py

start function and add walk rewrite (#14992)

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

device.py

compile QCOM without opening device (#15165)

2026-03-06 06:24:27 -05:00

dtype.py

cleanup fp8 conversion helpers and fp8 edge-case tests (#14953)

2026-02-22 09:16:42 -05:00

function.py

fix symbolic shape call + chunked prefill (#15149)

2026-03-05 14:02:26 +08:00

gradient.py

functions for llama trainer (#15045)

2026-02-28 12:15:18 +08:00

helpers.py

compile QCOM without opening device (#15165)

2026-03-06 06:24:27 -05:00

py.typed

add a single py.typed (#6083)

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

tensor.py

simplify image_conv2d pitch alignment hacks (#15158)

2026-03-05 07:17:34 -05:00