ggml-org/ggml

Pull Requests Commits

sync : llama.cpp

ggerganov committed 180 days ago

Verified 22d86019

ggml : Fix FP16 ELU positive branch (llama/16519)

sirus20x6 committed 180 days ago

Verified 99bb0226

ggml: Correct SVE implementation in ggml_vec_dot_f16_unroll (llama/16518)

sirus20x6 committed 180 days ago

Verified e5732875

CUDA: faster tile FA, add oob checks, more HSs (llama/16492)

JohannesGaessler committed 180 days ago

Verified 1077761d

sync : llama.cpp

ggerganov committed 181 days ago

Verified 9a238bf7

metal : fix mul-mm condition + fix mul-mv permuted kernels (llama/16494)

ggerganov committed 181 days ago

Verified a50be4ef

cuda : avoid initializing unused devices (llama/16510)

slaren committed 181 days ago

Verified 5c85a3c1

cmake : Dont define XOPENSOURCE on AIX (llama/16481)

mehendarkarprajwal committed 181 days ago

Verified d13a5fee

cpu : optimize the ggml NORM operation (llama/15953)

duduta committed 181 days ago

Verified c11d293e

CANN: Improve ACL graph matching (llama/16166)

noemotiovon committed 181 days ago

Verified 626d187f

kleidiai: kernel interface refactoring (llama/16460)

chaxu01 committed 181 days ago

Verified 91dbc935

refactor soft_max, add soft_max_back (llama/16472)

NeoZhangJianyu committed 181 days ago

Verified 77dd6c9e

Disable CUDA host buffers on integrated GPUs (llama/16308)

ai-fonsi committed 181 days ago

Verified b15f0b2e

metal : mark FA blocks (llama/16372)

ggerganov committed 181 days ago

Verified 5d3854a6

ggml webgpu: profiling, CI updates, reworking of command submission (llama/16452)

reeselevine committed 181 days ago

Verified 7255dc58

metal : add support for non-padded FA KV (llama/16148)

ggerganov committed 181 days ago

Verified d777d4a2

tests : add -INF blocks to the KQ mask in the FA tests (llama/16380)

ggerganov committed 181 days ago

Verified cf7fbdc1

metal : various optimizations + refactoring (llama/16446)

ggerganov committed 181 days ago

Verified 3d6002f2

ggml : fix unaligned access in AMX code (llama/16315)

ggerganov committed 181 days ago

Verified 53017b74

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

danbev committed 181 days ago

Verified a33af4d4

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

reeselevine committed 181 days ago

Verified 9bb0a5fe

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

netrunnereve committed 181 days ago

Verified 082230af

rpc : check src buffer when copying tensor (llama/16421)

rgerganov committed 181 days ago

Verified 0df2131b

rpc : add support for multiple devices (llama/16276)

rgerganov committed 181 days ago

Verified 4f0730d7

sync : llama.cpp

ggerganov committed 181 days ago

Verified e02c1345

vulkan : incremental shader builds (llama/16341)

Acly committed 181 days ago

Verified 2767954e

sync : llama.cpp

ggerganov committed 181 days ago

Verified 4a417818

metal : fix loop bound in ggml_mem_ranges (llama/16412)

ggerganov committed 181 days ago

Verified 5bd10f55

ggml : fix graph reallocation with multiple chunks (llama/16396)

Acly committed 181 days ago

Verified 187a56f0

vulkan: Replace uses of maxMemoryAllocationSize and VK_WHOLE_SIZE (llama/16354)

jeffbolznv committed 181 days ago

Verified f3fe64f1

Older