ggml-org/llama.cpp

Pull Requests Commits

minimal device tuning

0cc4m committed 1 day ago

bb254903

0cc4m committed 1 day ago

9403602b

vulkan: improve im2col memory write layout

0cc4m committed 2 days ago

a2a9ca8b

requirements : update transformers to 5.5.1 (#21617)

danbev committed 2 days ago

Verified c8ac02fa

webui: add "Send message on Enter" setting (#21577)

mourix committed 2 days ago

Verified 4ef9301e

common : fix ambiguous grammar rule in gemma4 (#21661)

aldehir committed 2 days ago

Verified ddf03c6d

common : simplify autoparser tagged parser rules (#21216)

aldehir committed 2 days ago

Verified 26229755

model: fix multimodal padding token for gemma3n/gemma4 (#21625)

ngxson committed 2 days ago

Verified 057dba33

mtmd: support dots.ocr (#17575)

ngxson committed 2 days ago

Verified 501aeed1

vocab: add gemma4 tokenizer tests, fix edge case (#21534)

pwilkin committed 2 days ago

Verified 0ec191e1

jinja : support ensure_ascii=true, string repetition and int/float self-filtering (#21623)

kwajiehao committed 2 days ago

Verified 243532e5

metal : add missing mm-id specializations for q1_0 (#21662)

ggerganov committed 2 days ago

Verified 5e9c6354

fix: Model Selector choice sync (#21628)

allozaur committed 2 days ago

Verified 9949ad08

server : fix grammar commandline args (#21543)

AUTOMATIC1111 committed 2 days ago

Verified 3ee9da0e

webui: Add option to pre-encode conversation for faster next turns (#21034)

allozaur committed 2 days ago

Verified 75511a8d

sycl : add flash-attn support for head size 512 (#21654)

qnixsynapse committed 2 days ago

Verified b54cb2e3

ci: drop v5 `all:` composition from labeler.yml (#21627)

Marxist-Leninist committed 2 days ago

Verified 8a65a7a8

vulkan: unify type macros to use Vx instead of _VECx (#21605)

0cc4m committed 2 days ago

Verified 8a132faa

common : skip non-primary GGUF split files when selecting model (#21633)

angt committed 2 days ago

Verified 42939190

CUDA: also store `node->src->data` ptrs for equality check (#21635)

am17an committed 2 days ago

Verified d12cc3d1

fix: free ctx_copy in ggml_opt_free to plug per-training-session leak (#21592)

RealOrko committed 2 days ago

Verified 2dcb7f74

server: respect the ignore eos flag (#21203)

ykhrustalev committed 2 days ago

Verified 66060008

vocab : remove </s> eog token if gemma4 (#21492)

aldehir committed 2 days ago

Verified d9a12c82

webui : send both backend_sampling == false/true (#18781)

ggerganov committed 2 days ago

Verified 4a05e0c5

Propose fix a couple of typos (#21581)

jeis4wpi committed 2 days ago

Verified e9fd9628

kv-cache : extend cache quantization checks (#21586)

Green-Sky committed 3 days ago

Verified 3ba12fed

webgpu : Query for adapter support when registering WebGPU backend (#21579)

reeselevine committed 3 days ago

Verified 54739490

metal: Q1_0 backend (#21528)

khosravipasha committed 3 days ago

Verified dcdcbad4

gemma : perform per-layer projections in the first layer (#21612)

ggerganov committed 3 days ago

Verified 5764d7c6

examples : disable cb_eval callback for --save-logits (#21553)

danbev committed 3 days ago

Verified 87f4744a

Older