Commits

History for

Commits on Jul 20, 2026

fix super long context loading issue
zhongkaifu
committed
47e6f67

Commits on Jul 18, 2026

improve cuda backend performance
zhongkaifu
committed
0fb0da7

Commits on Jul 9, 2026

fix Gemma 4 MTP error
zhongkaifu
committed
d691d7f

Commits on Jul 6, 2026

Merge branch 'main' into feat/cached-sampler
zhongkaifu
authored
d1b3496
refactor(runtime): simplify scheduler — remove dead code, fix live-cache race

Adrian
authored and
zhongkaifu
committed
75828e7
fix(runtime): cache TokenSampler per SequenceState, eliminate per-token LOH allocation
Adrian
committed
924f884

Commits on Jul 1, 2026

prefill optimization
zhongkaifu
committed
4631a25
update
zhongkaifu
committed
65dd2dd

Commits on Jun 29, 2026

optimize prefill
zhongkaifu
committed
a476458

Commits on Jun 28, 2026

fix KV cache reuse bug
zhongkaifu
committed
1a74206
optimize gemma decode
zhongkaifu
committed
64e8354

Commits on Jun 23, 2026

optimize Qwen3.6
zhongkaifu
committed
ab4e2a7

Commits on Jun 19, 2026

optimize mtp
zhongkaifu
committed
a012851

Commits on Jun 17, 2026

improve MTP performance on cuda backend
zhongkaifu
committed
2afee87
Run MTP speculative trunk through the batched paged path

zhongkaifu
and
claude
committed
f11de87
support Qwen3.6 mtp
zhongkaifu
committed
1ac7153

Commits on Jun 8, 2026

support gemma-4 uv and ua
zhongkaifu
committed
bc1295b
support gemma 4 uv
zhongkaifu
committed
a0a9abf

Commits on Jun 1, 2026

fix bugs
zhongkaifu
committed
72cd715

Commits on May 29, 2026

improve mlx backend performance
zhongkaifu
committed
20d4885

Commits on May 27, 2026

bug fix
zhongkaifu
committed
cd6ab84
support mlx backend
zhongkaifu
committed
687350f

Loading