ggerganov/llama.cpp

Pull Requests Commits

ci : refactor sdk caching to minimize storage (#16414)

CISC committed 3 days ago

Verified 3a002afa

ggml : fix unaligned access in AMX code (#16315)

ggerganov committed 4 days ago

Verified a23b9bdb

ci : remove missing reranker model files (#16444)

danbev committed 4 days ago

Verified 04e632a4

ggml-cpu : fix leftover handling in ggml_vec_scale_f32 for SVE (#16443)

danbev committed 4 days ago

Verified a80ff183

nix : removed metal for nix (#16118)

yuannan committed 4 days ago

Verified 1d49ca37

server: update readme to mention n_past_max metric (#16436)

okuvshynov committed 4 days ago

Verified c5fef0fc

model : Granite docling + Idefics3 preprocessing (SmolVLM) (#16206)

gabe-l-hart committed 5 days ago

Verified ca71fb9b

ggml webgpu: actually add softmax, fix rms_norm offset (#16400)

reeselevine committed 5 days ago

Verified 35266573

vulkan: use a more appropriate amount of threads when generating shaders (#16418)

netrunnereve committed 5 days ago

Verified 86df2c9a

rpc : check src buffer when copying tensor (#16421)

rgerganov committed 6 days ago

Verified f3928396

rpc : add support for multiple devices (#16276)

rgerganov committed 6 days ago

Verified 898acba6

vulkan : incremental shader builds (#16341)

Acly committed 6 days ago

Verified e29acf74

chat : support Magistral thinking (#16413)

ServeurpersoCom committed 6 days ago

Verified 128d522c

server : context checkpointing for hybrid and recurrent models (#16382)

ddh0 committed 6 days ago

Verified f6dcda39

metal : fix loop bound in ggml_mem_ranges (#16412)

ggerganov committed 6 days ago

Verified 606a73f5

llama : fix shapes for bert/mpt q/k norm (#16409)

CISC committed 7 days ago

Verified 946f71ed

ggml : fix graph reallocation with multiple chunks (#16396)

Acly committed 7 days ago

Verified 638d3302

Fix missing messages on sibling navigation (#16408)

allozaur committed 7 days ago

Verified 84c8e305

vulkan: Replace uses of maxMemoryAllocationSize and VK_WHOLE_SIZE (#16354)

jeffbolznv committed 7 days ago

Verified 2aaf0a2a

vulkan: Fix FA coopmat1 invalid array indexing (#16365)

jeffbolznv committed 7 days ago

Verified 0e1f8385

ci : change macos-13 to macos-15-intel (#16401)

danbev committed 7 days ago

Verified ad126479

Capture model name only after first token (streaming) or completed request (#16405)

allozaur committed 7 days ago

Verified 77233277

vulkan: in flash attention, bounds check against nem1 (don't rely on GGML_KQ_MASK_PAD) (#16316)

jeffbolznv committed 7 days ago

Verified e308efda

webui : Fix messages payload sent to chat completions (#16402)

allozaur committed 7 days ago

Verified 136bda78

fix: track viewportHeight via window.innerHeight to avoid unwanted scrolling (#16356)

ServeurpersoCom committed 7 days ago

Verified 5113efd3

test-barrier : do not use more threads than physically available (#16389)

CISC committed 7 days ago

Verified d64c8104

ggml webgpu: add support for soft_max, optimize rms_norm (#16357)

reeselevine committed 7 days ago

Verified ef07a409

model : Apertus model implementation (#15852)

pwilkin committed 7 days ago

Verified 34fcc5a4

musa: update compile flags (#16265)

yeahdongcn committed 7 days ago

Verified 91a2a565

ci : fix ubuntu-latest-cmake-rpc (disable ccache) (#16388)

CISC committed 8 days ago

Verified 72ee736c

Newer Older