[Bugfix] Tolerate an unset group_size in the Marlin MoE support probe - #17

Closed

afierka-intel wants to merge 15 commits into

mainfrom

afierka/marlin-utils-none-group-size

Commits on Aug 20, 2026

[Kernel][Kimi] fused vision q/k roper kernel (#50400)

lengrongfu
and
Isotr0py
authored
c8de519
[Bugfix] Return HTTP 400 instead of 501 for unknown chat roles in DeepSeek encoders (#53071)
JC-ut0
authored
38e9cef
[XPU][Tests] Make tests device-agnostic (#51968)
pmanczak
authored
44cf3f0
[ROCm] Fix DeepSeek V4 indexer numerics and coverage (#50803)
AndreasKaratzas
authored
6b68db4
[Bugfix] Record non-ImportError attention backend probe failures instead of crashing engine init (#51703)

Eoin-Houstoun
and
claude
authored
30e2394
[Rust Frontend] Fix Qwen parser auto-detection (#51169)

authored
727274a
[BugFix][Mooncake] Fix Mooncake saves from sparse Mamba block tables (#51362)
ZeldaHuang
authored
5b1e7a8
[CI] replace shellcheck script with shellcheck-py hook (#52572)

wjabbour
and
claude
authored
1eab6fe
[3/N] Harden Transformers modelling backend multi-modal path (#51827)
hmellor
authored
6e85feb
Fix weight tying (#51665)
hmellor
authored
f0c14b4
[Bugfix][GDN] Reset speculative decode count for an empty draft schedule (#53077)
khluu
authored
6df7adc
[Docs] Use incremental builds for C++ changes in `AGENTS.md` (#53098)

gau-nernst
and
codex
authored
6259572
[Model] Add tower and connector LoRA support for LFM2-VL (#51498)
zupengwang
authored
df13769
[Docker] Update to nixl-1.3.2 (#51777)

sandeep-maddipatla
and
jikunshang
authored
4b7cb94
[Bugfix] Normalise an unset group_size on the compressed-tensors WNA16 path

afierka-intel
and
claude
committed
cb59aed