ggerganov/llama.cpp

Pull Requests Commits

rm has_logit_bias

ngxson committed 37 days ago

a0f1a44b

address security issues

ngxson committed 37 days ago

bb9f5c08

llama: move suppress_tokens handling to common/sampling

ngxson committed 37 days ago

b1f192ff

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

KakaruHayate committed 37 days ago

Verified caa596ab

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

Titaniumtown committed 37 days ago

Verified 11b068d0

vendor: update BoringSSL to 0.20260728.0 (#26241)

cabelo committed 37 days ago

Verified e2f59ed7

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

ggerganov committed 37 days ago

Verified 992c3253

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

kaben committed 37 days ago

Verified e1af89a6

RPC: add tensor_memset (#25912)

am17an committed 37 days ago

Verified f5b9bd39

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

Geramy committed 37 days ago

Verified 60bccc37

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

satindergrewal committed 38 days ago

Verified 7be2c65d

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

crusaderky committed 38 days ago

Verified e9fa0781

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

reeselevine committed 38 days ago

Verified bc71c24c

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

wanghqc committed 38 days ago

Verified 8190848b

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

danbev committed 38 days ago

Verified 7e1e28ca

ui: rendering performance follow-up (#26097)

allozaur committed 38 days ago

Verified 6e2bc65f

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

jjerphan committed 38 days ago

Verified ad77bd31

server: abstract llama_memory calls to common_memory (#26221)

ngxson committed 38 days ago

Verified ee3d1b54

ggml : set output of view src (#25729)

am17an committed 38 days ago

Verified da5b4486

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

jeffbolznv committed 38 days ago

Verified 81616410

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

BLSharda committed 38 days ago

Verified b62b3509

spec: add DSpark speculative decoding (#25173)

wjinxu committed 38 days ago

Verified 84075273

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

aldehir committed 38 days ago

Verified 6ba5ef24

sycl: fix use-after-return of the SDPA scale in the oneDNN flash-attention path (#25880)

meatposes committed 38 days ago

Verified d6b61ac0

server : add extra trace log for prompt similarity (#26218)

ggerganov committed 38 days ago

Verified 9a3bf2b8

ggml-metal: FWHT kernel for metal backend (#25924)

nicklafleur committed 38 days ago

Verified f95de977

spec: add eagle3-v3 support for gpt-oss model (#25794)

ruixiang63 committed 38 days ago

Verified f8706784

contrib : add guideline about the "merge ready" label (#26178)

ggerganov committed 39 days ago

Verified c6292cfb

Disable -ffast-math on HIP (#25495)

Beinsezii committed 39 days ago

Verified 91f8c9c5

mtmd: support MiMo-V2.5 audio input (RVQ-based model) (#26190)

ngxson committed 39 days ago

Verified 1cbfd198

Older