ngxson/llama.cpp

Pull Requests Commits

kv-cache: follow the source cache size when sharing cells (#24267)

ServeurpersoCom committed 65 days ago

Verified f0156d14

llama : add Gemma4 MTP (#23398)

am17an committed 65 days ago

Verified 04eb4c44

spec : fix vocab compatibility check (#24256)

CISC committed 65 days ago

Verified 8a091c47

arg: Skip mmproj download when user supplied mmproj (#24239)

konradmb committed 65 days ago

Verified 465b1f0e

convert : fix Gemma4 with no audio encoder (#24242)

CISC committed 65 days ago

Verified f71af352

docker : bump cuda13 to 13.3.0 (#24228)

CISC committed 65 days ago

Verified 3f7c79d7

common/chat : fix LFM2/LFM2.5 reasoning round-trip and <think> leak (#24234)

tdakhran committed 65 days ago

Verified 98d5e8ba

mtmd: support "frame merge" for qwen-vl-based models (#21858)

ngxson committed 65 days ago

Verified 31e82494

completion : remove useless statics (#24226)

angt committed 66 days ago

Verified 6b80c74f

completion : fix format specifier in LOG_INF (#24213)

angt committed 66 days ago

Verified 588f0dc2

mtmd, server: add "placeholder bitmap" for counting tokens , add */input_tokens API (#23913)

ngxson committed 66 days ago

Verified f5c6ae18

vulkan: check coopmat2 features before reporting support (#24186)

0cc4m committed 66 days ago

Verified 5a69c974

model : rename local n_layer_all variable (#24209)

CISC committed 66 days ago

Verified 5343f450

context : fix off-by-one comparisons to n_gpu_layers (#24208)

CISC committed 66 days ago

Verified 603300b0

opencl: improve get_rows, cpy, concat and q6_k flat gemv (#24160)

lhez committed 66 days ago

Verified 308f61c3

common/chat : unify and fix LFM2/LFM2.5 tool parser (#24178)

tdakhran committed 66 days ago

Verified da87e9b6

vulkan: add fwht support for Intel with shmem reduction (#23964)

0cc4m committed 67 days ago

Verified e82beaa6

model: fix build failed (#24193)

ngxson committed 67 days ago

Verified c4a278d6

model, mtmd: Granite4 Vision (#23545)

gabe-l-hart committed 67 days ago

Verified 64086f2b

TP: round up granularity to 128 (#24180)

JohannesGaessler committed 67 days ago

Verified 6effcecd

cli: fix model params not propagated (#23893)

therealkenc committed 67 days ago

Verified 86591c75

model : fix llama_model::n_gpu_layers() (#24188)

ggerganov committed 67 days ago

Verified 96fbe003

ui: run npm install when package-lock.json is newer than node_modules (#24171)

ServeurpersoCom committed 67 days ago

Verified 2016bf2b

Fix link to available UI settings (#24169)

wariuccio committed 67 days ago

Verified 9c955c48

ui: add ignore-scripts=true to npmrc (#24149)

ngxson committed 67 days ago

Verified cc7bef34

docs: Update quantization readme (#24133)

pcuenca committed 67 days ago

Verified ad1b88ca

minor : fix lint issues (#24165)

ggerganov committed 67 days ago

Verified 59917d39

hparams : refactor `hparams.n_layer` (#24060)

ggerganov committed 67 days ago

Verified 7acb4e8c

kleidiai : dynamic chunck-based scheduling for hybrid execution (#23819)

chaxu01 committed 67 days ago

Verified 3ecfb150

CUDA: enroll mul_mat_vec_q_moe into pdl (#24087)

ORippler committed 67 days ago

Verified 2154a0fd

Newer Older