ngxson/llama.cpp

Pull Requests Commits

correct output_id for llama-cpp header

ngxson committed 1 year ago

c1f4a78f

fix embeddings and retrieval

ngxson committed 1 year ago

8ec0ff9b

server: fix batch_spec

ngxson committed 1 year ago

50fb3963

ngxson committed 1 year ago

56e82d02

ngxson committed 1 year ago

42062cc2

fix qwzn2vl mrope position input

ngxson committed 1 year ago

c4fea7fe

llama_batch_ext_init with ctx

ngxson committed 1 year ago

d18a79ed

Merge branch 'master' into xsn/private_batch_api

ngxson committed 1 year ago

1434c2c9

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

qwen2vl: fix mrope position

ngxson committed 1 year ago

a9efdbbc

Merge branch 'master' into xsn/private_batch_api

ngxson committed 1 year ago

e8827a6f

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

Older