chenyu da61088ca4 more divmod recombine (#15162)

2026-03-05 12:53:22 -05:00

..

apps

LLM speedup with two jits, prefill/rollout (#15153)

2026-03-05 16:21:09 +08:00

codegen

fully symbolic llm (#15097)

2026-03-05 10:22:11 +08:00

engine

fix symbolic shape call + chunked prefill (#15149)

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

mixin

move more to mixins (#14780)

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

nn

q5k is like q4k (#15155)

2026-03-05 17:02:49 +08:00

renderer

sqtt: simplify inst mapping, validate JUMP processing in CI (#15139)

2026-03-05 09:53:12 +09:00

runtime

Revert "bug in metal: offset is stored as uint32, overflow (#15129)" (#15136)

2026-03-04 16:54:42 -05:00

schedule

fix symbolic shape call + chunked prefill (#15149)

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

uop

more divmod recombine (#15162)

2026-03-05 12:53:22 -05:00

viz

viz: python traceback for USER device (#15160)

2026-03-05 20:22:09 +09:00

__init__.py

start function and add walk rewrite (#14992)

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

device.py

feat: per device mem_used (#15100)

2026-03-03 01:31:28 -08:00

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

LLM speedup with two jits, prefill/rollout (#15153)

2026-03-05 16:21:09 +08:00

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