llama.cpp
mtmd: support "frame merge" for qwen-vl-based models
#21858

Merged

mtmd: support "frame merge" for qwen-vl-based models #21858

ngxson merged 9 commits into master from mtmd-video-api

feat: add video support for Qwen3.5

573f2cf5

Merge branch 'master' into video-support

f558360b

various clean up

c5b682b2

github-actions added examples

Merge branch 'master' into mtmd-video-api

e5b3d6d1

Merge branch 'master' into mtmd-video-api

96e24ca5

revise the design

a404c4ea

ngxson changed the title ~~mtmd: support input sequence of images (initial video support)~~ mtmd: support "frame merge" for qwen-vl-based models 14 days ago

fix llava-uhd case

82b48212

nits

9819ad43

nits 2

b031b606

ngxson marked this pull request as ready for review 14 days ago

ngxson requested a review 14 days ago

ServeurpersoCom approved these changes on 2026-06-06

CISC approved these changes on 2026-06-06

ngxson merged 31e82494 into master 14 days ago

ngxson deleted the mtmd-video-api branch 7 days ago

Reviewers

CISC

ServeurpersoCom

Assignees

No one assigned

Labels

examples

Milestone

No milestone

llama.cpp mtmd: support "frame merge" for qwen-vl-based models #21858 Merged

mtmd: support "frame merge" for qwen-vl-based models #21858

llama.cpp
mtmd: support "frame merge" for qwen-vl-based models
#21858

Merged