ggml-org/llama.cpp

Pull Requests Commits

model: use ggml_rope_set_offset()

ngxson committed 7 hours ago

c993f31f

ggml: add ggml_rope_set_offset (+ metal support) (#27120)

ngxson committed 8 hours ago

Verified fe8156f7

ui: read persisted settings before the API key probe (#27365)

ServeurpersoCom committed 8 hours ago

Verified 77acca43

mtmd: add mtmd_bitmap_set_mergeable (#27348)

ngxson committed 8 hours ago

Verified 95c409c1

metal : dequantize q8_0 using packed types (#27370)

ggerganov committed 9 hours ago

Verified 8ef78e64

server: add dedup-cache-models preset option (#27346)

ngxson committed 11 hours ago

Verified ee4c505a

vulkan: tiled transpose for 0<->2 permuted CONT (#26585)

kh0pper committed 12 hours ago

Verified 98d1e92c

ggml-webgpu: add mulmat with overlapping src0/src1 (e.g., for minimax-01) (#27321)

yomaytk committed 13 hours ago

Verified 5112b973

ci : add attestation for signed release artifacts (#25933)

quartzjer committed 13 hours ago

Verified 0adcc3bb

tests: skip test-unicode build on win32/BUILD_SHARED_LIBS (#27336)

jeffbolznv committed 13 hours ago

Verified 79fe7991

gguf-py : add size guards to GGUFReader (#27188)

oakkaya committed 14 hours ago

Verified 0329fcda

server: (cosmetic) do not print cmd_child_to_router messages [no release] (#27347)

ngxson committed 20 hours ago

Verified 6d054983

opencl: fix WAR race in the generic FA tile kernels when the WG spans subgroups (#26434)

wanghqc committed 23 hours ago

Verified 60addddf

mtmd: deepseek-ocr SAM ggml_conv_2d with the im2col kept in F32 (#26727)

sfallah committed 1 day ago

Verified 9731ad3f

RPC: populate use_count to enable fusion inside backends (#27142)

am17an committed 1 day ago

Verified af517262

ui: Services consolidation refactor (#27239)

allozaur committed 1 day ago

Verified 3dc7285b

ui: Stores consolidation refactor (#27238)

allozaur committed 1 day ago

Verified fdf4c646

common: share thread pools when `n_threads` differ (#27138)

nikwen committed 1 day ago

Verified 04b56914

quant : Optimise memory usage by evicting weights after processing each layer (#22877)

EAddario committed 1 day ago

Verified 05967042

sycl: honor GGML_HINT_SRC0_IS_HADAMARD (#27298)

Titaniumtown committed 1 day ago

Verified 0882c7bc

unicode : include '~' in collapsed symbol class (#26972)

tarruda committed 1 day ago

Verified afd439df

readme : update status badges + regen AUTHORS (#27317)

ggerganov committed 1 day ago

Verified 169e4a7f

ci : Update OpenVINO to 2026.3, skip nemotron-h rollback test (#27292)

wine99 committed 1 day ago

Verified 9d77fa17

ci: add Windows ARM64 CUDA support to the manual workflow (#27300)

shivamkumard-ctrl committed 1 day ago

Verified c0296022

mtmd: fix LFM2 image tiling threshold (#27057)

BlackFoil committed 1 day ago

Verified 7acdbb1f

ggerganov committed 1 day ago

1511ce3b

ggml : bump version to 0.20.2 (ggml/1589)

ggerganov committed 1 day ago

da786dc2

build : fix xcframework + cmake clean-up (#27304)

ggerganov committed 1 day ago

Verified 27e345b5

ci : create pre-release with change log and nightly link in make-release (#27302)

ggerganov committed 1 day ago

Verified 8b864009

CUDA: MMVQ nwarps=8 for bs=1 for dense models on DGX Spark (#26843)

ynankani committed 1 day ago

Verified 25ae3a9b

Older