ggml-org/llama.cpp

Pull Requests Commits

ggml: generalize the k-pool mask op into ggml_top_k_mask

ServeurpersoCom committed 3 days ago

55672a51

tests : initialize the remaining tensors in test_kpool_mask

am17an committed 3 days ago

2961d5e5

enroll glm5.3-next too

am17an committed 5 days ago

6459705e

llama: create new fused op for QSA scoring

am17an committed 5 days ago

4ac70767

tests : update Metal fusion baseline for beam (#30764)

ggerganov committed 5 days ago

Verified 27af5816

pi : turn .pi/gg into a system prompt addendum (#30762)

ggerganov committed 5 days ago

Verified 7db0d3d7

CUDA: fix shared-memory WAR hazard in flash_attn_ext_vec (#30573)

invaliddayta committed 5 days ago

Verified 99db719b

hexagon: FLASH_ATTN_EXT improvements for longer context and SWA models (#30731)

max-krasnyansky committed 5 days ago

Verified 31288c9c

hexagon: allow multistream FA (#30218)

jhen0409 committed 5 days ago

Verified fe8841a7

cuda : fuse qwen4exp's SCALE -> SIGMOID -> SCALE -> hc_post chain (#30602)

am17an committed 6 days ago

Verified ff5a2337

ggml: support non-contiguous SUM (CPU, CUDA, Vulkan, WebGPU) (#28851)

rssndr committed 6 days ago

Verified 0878f7c3

ggml-cuda: fattn do not check oob if k_VKQ_max = (k_VKQ_max - nbatch_fa) (#30572)

thelittlefireman committed 6 days ago

Verified 5aeac37a

Restore ESIMD kernels on FFN path (#30239)

malsbat committed 6 days ago

Verified 547208dc

hexagon: add support for hadamard MUL_MAT and GET_ROW for flat and tiled Q1_0 and Q2_0 (#30627)

tboinovski1 committed 6 days ago

Verified 848bbda9

hexagon: add support for Q1_0 and Q2_0 (#30122)

tboinovski1 committed 6 days ago

Verified 797de63b

ggml-cuda: HIP: use __builtin_amdgcn_perm for q1_0 unpacking (#29927)

thelittlefireman committed 7 days ago

Verified 9d1878a4

ui : Discover & Download Models from HF Hub (#29584)

allozaur committed 7 days ago

Verified c0fd7862

CUDA: faster MMVQ for low-bit types (#30517)

am17an committed 7 days ago

Verified 2744e48d

tests: give the random model generator a per-tensor filler table (#30674)

ServeurpersoCom committed 7 days ago

Verified 10b8b755

cuda : fix -Wtautological-compare in fattn vec dispatch (#30519)

angt committed 7 days ago

Verified 9d1e0361

cuda : add grouped experts top-k fusion (#29181)

CISC committed 7 days ago

Verified f2f667b0

model : add Kolibri-1 (Aleph-Alpha) support (#30673)

riteme committed 7 days ago

Verified 62318cd5

ci : added ABI reports as github artifact (#30619)

sandrohanea committed 7 days ago

Verified 83f93894

model : Granite switch decision 256 options (#30693)

gabe-l-hart committed 7 days ago

Verified 1268edac

sycl: fix prompt-processing regression after Q4_0/Q8_0 weight reorder (#30374)

marcusds committed 7 days ago

Verified f74e2fa6

skills: add slice-model (#30677)

ngxson committed 7 days ago

Verified ea0c4c2c

ci: update device matrix for Snapdragon jobs by removing SM8750 (#30620)

zhiyuan8 committed 7 days ago

Verified 6420cf5d

hexagon: vectorize softplus with HVX (#30435)

huangwei858 committed 7 days ago

Verified 27c7a769

opencl: Add MxFP4 MoE bin kernel for Adreno X1 devices (#30691)

shawngu-quic committed 7 days ago

Verified 2f2763db

model: add Beam (Reflection) (#30670)

danielhanchen committed 7 days ago

Verified 01fc3aae

Older