ggml-org/llama.cpp

Pull Requests Commits

ngxson committed 1 day ago

7bc6c2b8

Merge branch 'master' into xsn/mtmd_audio_stream

ngxson committed 1 day ago

0c3a08f5

model: add Muse Spark Open (text, vision and audio) (#30616)

ServeurpersoCom committed 1 day ago

Verified ba3c05f7

mtmd: add mtmd_audio_stream + support voxtral-realtime

ngxson committed 1 day ago

46f6b608

pi : expand AI disclosure with hardware and stats (#30615)

ggerganov committed 1 day ago

Verified b76b1337

llama.cpp : bump version to 0.7.0 (#30603)

ggerganov committed 2 days ago

Verified 620d2225

models : add Nemotron Super 3.5 VL and Nemotron Bolt 74B A9B support (#30520)

frozenblade1224 committed 2 days ago

Verified cd0a5b6a

scripts : add ci-status.sh to list active CI runs (#30586)

ggerganov committed 2 days ago

Verified a11b6530

server: fix the off-by-two in the SWA reuse threshold (#30580)

ServeurpersoCom committed 2 days ago

Verified 61ad80bf

model : Mellum - read MoE gating params from GGUF, unify RoPE (#30578)

Xarbirus committed 2 days ago

Verified a6b91d8c

server: save the draft state in the slot file (#30574)

ServeurpersoCom committed 2 days ago

Verified ea736c1d

server: prefer empty slots and keep states overwritten by child tasks (#30567)

ServeurpersoCom committed 2 days ago

Verified b0d23916

ggerganov committed 2 days ago

c0cb53ac

ggml : bump version to 0.27.0

ggerganov committed 2 days ago

4ed1c850

ci : bump ty to 0.0.88 (#30587)

CISC committed 2 days ago

Verified 711780ff

fix(ui): take the agentic turns default from DEFAULT_AGENTIC_CONFIG (#30582)

allozaur committed 2 days ago

Verified fd55cb98

models : add maion-coder (#29778)

mailkv23 committed 2 days ago

Verified 3e4dfa8f

jinja : improve context copy constructor performance (#30562)

Ninja3047 committed 2 days ago

Verified 08698c81

jinja : write nan and inf in tojson as json.dumps does (#30483)

ykhrustalev committed 2 days ago

Verified 82e1e2fd

vulkan : fix concurrent first device init (#29766)

losewayy committed 2 days ago

Verified 47f5a64a

sycl : replace oneAPI package by DLE+oneDNN package (#30131)

arthw committed 2 days ago

Verified e36df604

opencl: fix OOB tail access for adreno noshuffle gemv kernels (#30421)

lhez committed 2 days ago

Verified 9166ceaf

jinja : fix loop body variable scoping (#30533)

Ninja3047 committed 2 days ago

Verified 2130ca84

ggml-cpu : optimize pad_f32 (#30526)

angt committed 2 days ago

Verified 23f83e05

unicode : reject overlong UTF-8 and out-of-range codepoints (#30485)

simoneiacomino committed 2 days ago

Verified 2ce20d22

llama : split llama_prefetch() into merge and prefetch steps (#30532)

angt committed 2 days ago

Verified 68ac7657

opencl: apply flash attn softcap before mask (#30546)

lhez committed 2 days ago

Verified 77c1f399

vulkan : copy strided f16 KV once on AMD so reads spread across memory channels (#27703)

Nathanw1014 committed 2 days ago

Verified 733f57a9

cpu: add F16 SWIGLU_OAI reference implementation and validation (#29200)

cqderek committed 2 days ago

Verified 5a638aed

metal : add BF16 variant of the tensor API flash attention kernel (#30491)

ethanhq committed 2 days ago

Verified 6e66cb5d

Older