PR #27941 qwen4exp: follow up fixes

qwen4exp: follow up fixes #27941

ggerganov merged 8 commits into ggml-org:master from danielhanchen:qwen4exp/followup-fixes-squashed

github-actions added model

Bombocloot approved these changes on 2026-08-30

danielhanchen marked this pull request as ready for review 8 days ago

danielhanchen requested a review from

CISC 8 days ago

danielhanchen requested a review from

ggerganov 8 days ago

ggerganov commented on 2026-08-30

ggerganov commented on 2026-08-31

github-actions added testing

danielhanchen force pushed from 02eb201d to 6b2b85cf 7 days ago

qwen4exp: follow up fixes

3dd55057

-kvu NaN collapse fix

4cbf055d

indexer cache ext.x/ext.y restore fix

9bfa91ce

kv-cells: rename seq_set to seq_get_all

c76ea01d

memory-hybrid-idx: implement set_input_qsa on the memory class

4ebbc37d

tests: check that a sequence state survives a save/restore round-trip

52bc9ff1

danielhanchen force pushed from 6b2b85cf to 52bc9ff1 7 days ago

tests: give the synthetic qwen4exp a PLE so the state test bites

21c13c48

danielhanchen requested a review from

JohannesGaessler 7 days ago

ggerganov commented on 2026-09-01

llama: disable -sm tensor for qwen4exp

c961cd31

ServeurpersoCom approved these changes on 2026-09-01

ggerganov approved these changes on 2026-09-01

ggerganov merged 36b10154 into master 6 days ago

Reviewers

ggerganov

ServeurpersoCom

Bombocloot

CISC

JohannesGaessler

Assignees

No one assigned

Labels

model testing

Milestone

No milestone

llama.cpp qwen4exp: follow up fixes #27941 Merged

qwen4exp: follow up fixes #27941

llama.cpp
qwen4exp: follow up fixes
#27941

Merged