ggml-org/llama.cpp

Pull Requests Commits

ngxson committed 4 hours ago

e37f8c8e

add --video arg

ngxson committed 4 hours ago

ed79fd1b

cli: allow auto-completion for video

ngxson committed 5 hours ago

c787220f

ngxson committed 5 hours ago

a9dff818

ngxson committed 5 hours ago

677828cf

add MTMD_VIDEO config

ngxson committed 5 hours ago

bf417027

support video input on server (base64 input)

ngxson committed 6 hours ago

24d72575

add mtmd_helper_video

ngxson committed 6 hours ago

570a4ef8

ngxson committed 6 hours ago

1361372d

remember to free lazy text

ngxson committed 8 hours ago

16e74d64

ok: lazy bitmap API

ngxson committed 8 hours ago

5375d693

Merge branch 'master' into xsn/mtmd-helper-video-input

ngxson committed 10 hours ago

20a06e9d

ngxson committed 11 hours ago

aa687d7a

arg: Skip mmproj download when user supplied mmproj (#24239)

konradmb committed 12 hours ago

Verified 465b1f0e

convert : fix Gemma4 with no audio encoder (#24242)

CISC committed 14 hours ago

Verified f71af352

docker : bump cuda13 to 13.3.0 (#24228)

CISC committed 15 hours ago

Verified 3f7c79d7

common/chat : fix LFM2/LFM2.5 reasoning round-trip and <think> leak (#24234)

tdakhran committed 1 day ago

Verified 98d5e8ba

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

ngxson committed 1 day ago

Verified 31e82494

completion : remove useless statics (#24226)

angt committed 1 day ago

Verified 6b80c74f

completion : fix format specifier in LOG_INF (#24213)

angt committed 1 day ago

Verified 588f0dc2

mtmd, server: add "placeholder bitmap" for counting tokens , add */input_tokens API (#23913)

ngxson committed 1 day ago

Verified f5c6ae18

vulkan: check coopmat2 features before reporting support (#24186)

0cc4m committed 1 day ago

Verified 5a69c974

model : rename local n_layer_all variable (#24209)

CISC committed 1 day ago

Verified 5343f450

context : fix off-by-one comparisons to n_gpu_layers (#24208)

CISC committed 1 day ago

Verified 603300b0

opencl: improve get_rows, cpy, concat and q6_k flat gemv (#24160)

lhez committed 2 days ago

Verified 308f61c3

common/chat : unify and fix LFM2/LFM2.5 tool parser (#24178)

tdakhran committed 2 days ago

Verified da87e9b6

vulkan: add fwht support for Intel with shmem reduction (#23964)

0cc4m committed 2 days ago

Verified e82beaa6

model: fix build failed (#24193)

ngxson committed 2 days ago

Verified c4a278d6

model, mtmd: Granite4 Vision (#23545)

gabe-l-hart committed 2 days ago

Verified 64086f2b

TP: round up granularity to 128 (#24180)

JohannesGaessler committed 2 days ago

Verified 6effcecd

Older