ggml-org/llama.cpp

Pull Requests Commits

fix ~server_batch()

ngxson committed 73 days ago

76bb9cb0

server: support embd for sampled token

ngxson committed 74 days ago

d8625645

Remove custom cpu op from the M3 graph, express with stock ops (#26297)

timkhronos committed 74 days ago

Verified b2f22168

metal: fix memory unwire if model is freed without any GPU operations (#26082)

nikwen committed 74 days ago

Verified d0bfb198

ui: IndexedDB and Conversations data fixes (#26278)

allozaur committed 74 days ago

Verified 21a5f5b7

ggml : Fix issue with kleidiai ci and stringop overflow warning (#26277)

JonathanC-ARM committed 74 days ago

Verified 32703b42

[UT] enhance UT to show all real unsupported backends (#25234)

arthw committed 74 days ago

Verified a6a77bc4

mimo2: address MTP review feedback (#26228)

tnhnyzc committed 74 days ago

Verified 64d528be

fix: increase greeting spacing on md screens (#26287)

allozaur committed 74 days ago

Verified 3018a11e

llama: move suppress_tokens handling to common/sampling (#26276)

ngxson committed 74 days ago

Verified afeebe10

ggml-cuda : disable MMQ on devices with less than 48 KiB shared memory (#26141)

KakaruHayate committed 75 days ago

Verified caa596ab

sycl: contiguous fast path + 32-bit index math for unary elementwise ops (#25946)

Titaniumtown committed 75 days ago

Verified 11b068d0

vendor: update BoringSSL to 0.20260728.0 (#26241)

cabelo committed 75 days ago

Verified e2f59ed7

server : add trace logging for slot similarity checking (#26271)

ggerganov committed 75 days ago

Verified 992c3253

conversion: fix Qwen2.5-Omni mmproj conversion regression (#26262)

kaben committed 75 days ago

Verified e1af89a6

RPC: add tensor_memset (#25912)

am17an committed 75 days ago

Verified f5b9bd39

add rdna3.5, and 3 to mmq configs so they can be tuned independently. (#26199)

Geramy committed 75 days ago

Verified 60bccc37

model: add NextN/MTP speculative decoding support for GLM_DSA (GLM-5.2) (#25980)

satindergrewal committed 75 days ago

Verified 7be2c65d

model: Add Laguna-S-2.1 LLM_TYPE (#26233)

crusaderky committed 75 days ago

Verified e9fa0781

ggml-webgpu: Fix some binding alias issues to support all archs, fix recurrent-state-rollback test (#25931)

reeselevine committed 75 days ago

Verified bc71c24c

opencl: skip the Adreno KQ/KQV image kernels for multi-stream batches (#26189)

wanghqc committed 75 days ago

Verified 8190848b

mtmd : add Nemotron 3 Nano Omni support (parakeet) (#22520)

danbev committed 75 days ago

Verified 7e1e28ca

ui: rendering performance follow-up (#26097)

allozaur committed 75 days ago

Verified 6e2bc65f

docs: Adapt conda-forge package name (#26229)

jjerphan committed 75 days ago

Verified ad77bd31

server: abstract llama_memory calls to common_memory (#26221)

ngxson committed 75 days ago

Verified ee3d1b54

ggml : set output of view src (#25729)

am17an committed 75 days ago

Verified da5b4486

vulkan: add iq4_nl support back to FA (#24585)

jeffbolznv committed 76 days ago

Verified 81616410

ggml-cuda: add chunked SSD matmul for Mamba-2 prefill acceleration (#22675)

BLSharda committed 76 days ago

Verified b62b3509

spec: add DSpark speculative decoding (#25173)

wjinxu committed 76 days ago

Verified 84075273

common/chat: add specialized minimax m3 parser (#26210)

aldehir committed 76 days ago

Verified 6ba5ef24

Older