ggml-org/llama.cpp

Pull Requests Commits

convert : match ssm_conv tensors by type

compilade committed 1 year ago

20b256e0

convert : fix squeeze for ssm_conv tensors

ggerganov committed 1 year ago

Verified 9c60fc4c

ggml-cpu : update KleidiAI to v1.5.0 (#12568)

eddnjjn committed 1 year ago

Verified 053b3f9a

SYCL: disable Q4_0 reorder optimization (#12560)

qnixsynapse committed 1 year ago

Verified e2f56017

docs : add build instructions for KleidiAI (#12563)

eddnjjn committed 1 year ago

Verified 36ee06dd

ci: [MUSA] add CI and update doc (#12562)

yeahdongcn committed 1 year ago

Verified 3cd3a395

context : fix worst-case reserve outputs (#12545)

ggerganov committed 1 year ago

Verified 2d77d88e

ci: [SYCL] ggml-ci Use main GPU and enable sysman (#12547)

qnixsynapse committed 1 year ago

Verified c95fa362

opencl: simplify kernel embedding logic in cmakefile (#12503)

lhez committed 1 year ago

Verified 2b65ae30

CI: fix SYCL build (#12546)

qnixsynapse committed 1 year ago

Verified 48d7021c

docs: update: improve the Fedoa CUDA guide (#12536)

teihome committed 1 year ago

Verified 3361e2de

llama-vocab : add SuperBPE pre-tokenizer (#12532)

compilade committed 1 year ago

Verified 00d53800

CUDA: Fix clang warnings (#12540)

yeahdongcn committed 1 year ago

Verified 7ea75035

mmap : skip resource limit checks on AIX (#12541)

mehendarkarprajwal committed 1 year ago

Verified c54f6b79

vulkan: fix mul_mat_vec failure in backend tests (#12529)

jeffbolznv committed 1 year ago

Verified 9b169a4d

server : Add verbose output to OAI compatible chat endpoint. (#12246)

mglambda committed 1 year ago

Verified 77f9c6bb

install : add macports (#12518)

IOOI-SqAR committed 1 year ago

Verified 18b663d8

llama : gemma3 : use output tensor if it exists in model weight (#12506)

ngxson committed 1 year ago

Verified fbdfefe7

ggml : fix quantized cpy op (#12310)

ggerganov committed 1 year ago

Verified ba932dfb

musa: refine compute capability (#12493)

yeahdongcn committed 1 year ago

Verified fac63a3d

vulkan: Optimize mul_mat_vec p021 and nc shaders (#12505)

jeffbolznv committed 1 year ago

Verified eddfb438

Vulkan: RTE rounding for cpy to quant (#12480)

stduhpf committed 1 year ago

Verified 4375415b

vulkan: workaround for AMD Windows driver 16 bit unpack8 bug (#12472)

netrunnereve committed 1 year ago

Verified 30c42ef5

model : do not repack if a GPU device is present (#12498)

ggerganov committed 1 year ago

Verified af04481e

chore : cleanup llama_model_loader::TENSOR_ usage (#12492)

CISC committed 1 year ago

Verified 960e7260

llama-tts : avoid crashes related to bad model file paths (#12482)

marcoStocchi committed 1 year ago

Verified ea1518e8

[SYCL] Fix build on Windows when ccache enabled (#9954) (#9976)

MakeDecisionWorth committed 1 year ago

Verified 1aa87ee5

sycl: cleanup oneDNN related code (#12097)

sgeor255 committed 1 year ago

Verified 9ffcc9e3

webui : Prevent rerendering on textarea input (#12299)

woof-dog committed 1 year ago

Verified e0464306

llama : make Qwen2MoE QKV bias optional (#12477)

CISC committed 1 year ago

Verified dbb3a473

Older