ggerganov/llama.cpp

Pull Requests Commits

sycl : variable sg_size support for mmvq kernels (#12336)

Alcpz committed 212 days ago

Verified 363f8c5d

CUDA/HIP: Fix fattn-vec-* when device warp size is not 32 (#12315)

IMbackK committed 212 days ago

Verified 34c961b1

llama : Add Gemma 3 support (+ experimental vision capability) (#12343)

ngxson committed 212 days ago

Verified 7841fc72

vulkan: fix bug in coopmat1 mul_mat_id (#12316)

jeffbolznv committed 212 days ago

Verified bf69cfe6

CUDA/HIP: refractor mmqv to unify the calculation of nwarps and rows per block between host and device code. (#12177)

IMbackK committed 213 days ago

Verified 10f2e818

ggml-backend : fix backend search path (#12330)

jklincn committed 213 days ago

Verified ba765438

metal : Cache the Metal library at the device context level (#12265)

BB-fat committed 213 days ago

Verified 6ab2e476

clip : bring back GPU support (#12322)

ngxson committed 213 days ago

Verified 96e12808

mat vec double buffer (#12188)

netrunnereve committed 214 days ago

Verified 2c9f833d

musa: support new arch mp_31 and update doc (#12296)

yeahdongcn committed 214 days ago

Verified 25136454

opencl: use OpenCL C standard supported by the device (#12221)

Henry Linjamäki committed 214 days ago

Verified 8acdacb3

readme: added Sidekick to available UIs (#12311)

johnbean393 committed 214 days ago

Verified 89b2b56e

tests : fix test-quantize-fns to init the CPU backend (#12306)

ggerganov committed 214 days ago

Verified e128a1bf

common : refactor '-o' option (#12278)

marcoStocchi committed 214 days ago

Verified 6ef79a67

`server`: extract <think> tags from qwq outputs (#12297)

ochafik committed 214 days ago

Verified 4e39a3c3

`tool-call`: ensure there's always a non-empty tool call id (#12292)

ochafik committed 214 days ago

Verified be421fc4

allow missing content in message if tool_calls provided (#12293)

ochafik committed 214 days ago

Verified 87c26305

`sampler`: fixes trigger tokens + lazy grammars (fix typo cast from token to string) (#12291)

ochafik committed 214 days ago

Verified 2b3a25c2

llava : fix bug in minicpm-v code (#11513)

tc-mb committed 214 days ago

Verified 8352cdc8

server : add speculative decoding presets for FIM (#12287)

ggerganov committed 215 days ago

Verified 1e2f78a0

authors : update (#12271)

ggerganov committed 216 days ago

Verified 0fd7ca7a

ggml-backend : make path_str compatible with C++20 (#12269)

ctrysbita committed 216 days ago

Verified 6fefc05a

server : infill gen ends on new line (#12254)

ggerganov committed 217 days ago

Verified 7ab36439

ggml : skip intermediate .air file when compiling .metallib (#12247)

danbev committed 217 days ago

Verified 7c7f3b7f

ggerganov committed 217 days ago

102ac189

ggml : ggml_compute_forward_concat() for arbitrary tensor type (ggml/1118)

vmobilis committed 217 days ago

d6ae2fa0

ggml-cpu: faster AVX2 variant for IQ1_M (#12216)

remyoudompheng committed 217 days ago

Verified 68d0027f

ci : fix save-load test invocations (#12245)

ggerganov committed 217 days ago

Verified ea002810

server : Log original chat template parsing error (#12233)

CISC committed 217 days ago

Verified 8fad3c7a

sync: minja - support QwQ-32B (#12235)

ochafik committed 217 days ago

Verified 7cf64f6b

Newer Older