ggerganov/llama.cpp

Pull Requests Commits

Make logging more verbose (#11714)

ericcurtin committed 271 days ago

Verified d2fe216f

llama : fix defrag logic (#11707)

ggerganov committed 271 days ago

Verified ed926d88

vocab : ignore invalid UTF-8 input in the BPE tokenizer (#11729)

cfillion committed 271 days ago

Verified 2d219b38

llama : fix progress dots (#11730)

magicse committed 271 days ago

Verified 333820d7

vulkan: print shared memory size (#11719)

jeffbolznv committed 271 days ago

Verified c026ba3c

llama : add llama_sampler_init for safe usage of llama_sampler_free (#11727)

cfillion committed 271 days ago

Verified 7ee953a6

SYCL: remove XMX info from print devices (#11712)

qnixsynapse committed 271 days ago

Verified ec3bc827

common : add default embeddings presets (#11677)

danbev committed 271 days ago

Verified b7552cfc

ggml : optimize and build warning fix for LoongArch (#11709)

MQ-mengqing committed 271 days ago

Verified 225bbbfa

llama : fix old glm4 models (#11670)

tv1wnd committed 272 days ago

Verified 855cd073

ggerganov committed 272 days ago

Verified 8a59053f

rpc: fix known RCE in rpc-server (ggml/1103)

retr0reg committed 272 days ago

Verified 1d20e53c

server : (webui) migrate project to ReactJS with typescript (#11688)

ngxson committed 272 days ago

Verified 2fb3c32a

docs: update fedora cuda guide for 12.8 release (#11393)

teihome committed 272 days ago

Verified 9ab42dc7

SYCL: Adjust support condition for norm operators (#11674)

qnixsynapse committed 272 days ago

Verified 194b2e69

llama : add log about loading model tensors (#11699)

ggerganov committed 272 days ago

Verified 9dd7a039

build : fix llama.pc (#11658)

angt committed 272 days ago

Verified c0d48432

ggml : fix LoongArch compile error with 128-bit SIMD (#11701)

junchao-loongson committed 272 days ago

Verified 8d4d2be1

vulkan: optimize coopmat2 iq2/iq3 callbacks (#11521)

jeffbolznv committed 273 days ago

Verified 2c6c8df5

vulkan: initial support for IQ4_XS quantization (#11501)

remyoudompheng committed 273 days ago

Verified 8a7e3bf1

vulkan: use smaller combined allocations to avoid fragmentation (#11551)

jeffbolznv committed 273 days ago

Verified 1b598b30

metal : avoid breaking build when metal API predates TARGET_OS_VISION (#11690)

charles-dyfis-net committed 273 days ago

Verified 902368a0

readme : add link to Autopen under UIs (#11684)

blackhole89 committed 273 days ago

Verified c3db0480

metal : adjust support conditions for norm operators (#11671)

ggerganov committed 273 days ago

Verified d774ab3a

CUDA: support for mat. mul. with ne03 != ne13 (#11656)

JohannesGaessler committed 273 days ago

Verified fa62da9b

llava: add quantization for the visual projector LLAVA, Qwen2VL (#11644)

samkoesnadi committed 273 days ago

Verified 1ec20808

`sync`: minja (#11641)

ochafik committed 274 days ago

Verified 9f4cc8f8

CUDA: non-contiguous (RMS) norm support (#11659)

JohannesGaessler committed 274 days ago

Verified fd08255d

HIP: force max threads per block to be 1024 (#11621)

fxzjshm committed 274 days ago

Verified 3ec9fd4b

server : add try..catch to places not covered by set_exception_handler (#11620)

ngxson committed 274 days ago

Verified 3962fc1a

Newer Older