ngxson/llama.cpp

Pull Requests Commits

rm outdated comments

ngxson committed 49 days ago

06a5ebe1

trailing whitespace

ngxson committed 49 days ago

b51da9a1

ngxson committed 49 days ago

1663d2f8

fix data race on accessing props and model_meta

ngxson committed 49 days ago

d8500827

don't allow accessing /models during sleep, it is not thread-safe

ngxson committed 49 days ago

0bb9bc48

allow /models to be accessed during sleeping, fix use-after-free

ngxson committed 49 days ago

fd09f880

fix use-after-free

ngxson committed 49 days ago

105e2f3c

ngxson committed 49 days ago

937b0641

make sure to free llama_init

ngxson committed 49 days ago

e6ab62c4

optimization: add fast path

ngxson committed 49 days ago

44a5a26c

ngxson committed 49 days ago

aea8f8c1

ngxson committed 49 days ago

db3b78d2

implement server-context suspend

ngxson committed 49 days ago

197e5785

implement sleeping at queue level

ngxson committed 49 days ago

e1d7b434

server : [easy] fix per round speculative decode logging (#18211)

okuvshynov committed 50 days ago

Verified 408616ad

server: support load model on startup, support preset-only options (#18206)

ngxson committed 50 days ago

Verified 9e39a1e6

ci : remove non-windows zip artifacts (#18201)

CISC committed 50 days ago

Verified 74e05131

ci : only save ccache on master (#18207)

CISC committed 50 days ago

Verified f74747d8

ggml-hexagon: Implement true Q8_0 quantization on Hexagon NPU for more accurate mixed-precision matmul operations (#17977)

ngdxzy committed 50 days ago

Verified ce734a8a

arg: fix order to use short form before long form (#18196)

ServeurpersoCom committed 50 days ago

Verified 14931a82

llama : Changing off_t to size_t for Windows (#18204)

JTischbein committed 50 days ago

Verified f99ef53d

server: friendlier error msg when ctx < input (#18174)

am17an committed 51 days ago

Verified cc0a0434

presets: refactor, allow cascade presets from different sources, add global section (#18169)

ngxson committed 51 days ago

Verified 98c1c7a7

webui: Add editing attachments in user messages (#18147)

allozaur committed 51 days ago

Verified acb73d83

model-conversion : add verbose flag in run-org-model.py (#18194)

danbev committed 51 days ago

Verified 0a271d82

android: fix missing screenshots for Android.md (#18156)

naco-siren committed 51 days ago

Verified 52fc7fee

vulkan: Add perf logger mode with concurrency (#17944)

jeffbolznv committed 51 days ago

Verified cdbada8d

model : add ASR support for LFM2-Audio-1.5B (conformer) (#18106)

ngxson committed 51 days ago

Verified 8ea958d4

webui: display prompt processing stats (#18146)

ServeurpersoCom committed 51 days ago

Verified f9ec8858

ggml-cpu: extend support for RVV floating-point kernels (#17318)

taimur-10x committed 51 days ago

Verified f716588e

Older