ngxson/llama.cpp

Pull Requests Commits

mtmd: allow QwenVL to process larger image by default

ngxson committed 52 days ago

bc169a26

ci : apply model label to models (#16994)

CISC committed 53 days ago

Verified d9458343

chore : fix models indent after refactor (#16992)

CISC committed 53 days ago

Verified b164259b

Fix garbled output with REPACK at high thread counts (#16956)

NoahOksuz committed 53 days ago

Verified 1f5accb8

CUDA: avoid mul + bias fusion when doing fusion (#16935)

am17an committed 53 days ago

Verified 2759ccdb

opencl: support imrope (#16914)

lhez committed 53 days ago

Verified c5023daf

fix: Viewing multiple PDF attachments (#16974)

allozaur committed 54 days ago

Verified e7da30b5

model-conversion : pass config to from_pretrained (#16963)

danbev committed 54 days ago

Verified ed8aa633

server : add props.model_alias (#16943)

ggerganov committed 54 days ago

Verified 48bd2650

ggml: CUDA: add head size 72 for flash-attn (#16962)

theo77186 committed 54 days ago

Verified 622cd010

mtmd: add --image-min/max-tokens (#16921)

ngxson committed 54 days ago

Verified 070ff4d5

mtmd: pad mask for qwen2.5vl (#16954)

ngxson committed 54 days ago

Verified bf7b0c97

ggml : LoongArch fixes (#16958)

MQ-mengqing committed 54 days ago

Verified fcfce040

sync: minja (glm 4.6 & minmax m2 templates) (#16949)

ochafik committed 54 days ago

Verified ee3a5a10

SYCL: optimized repeat_back kernel (3× fewer asm instructions, 2× faster)Feature/sycl repeat back opt (#16869)

shani-f committed 54 days ago

Verified 7e994168

feat(webui): improve LaTeX rendering with currency detection (#16508)

srogmann committed 54 days ago

Verified bcfa8762

test-backend-ops : fix segfault in moe-expert-reduce test in support mode and coverage (#16936)

sbera77 committed 54 days ago

Verified a2054e3a

ci : disable failing riscv cross build (#16952)

CISC committed 54 days ago

Verified dd528680

model: add Janus Pro for image understanding (#16906)

ravenouse committed 54 days ago

Verified 6b9a5242

clip : use FA (#16837)

ggerganov committed 54 days ago

Verified 2f966b8e

server : support unified cache across slots (#16736)

ggerganov committed 55 days ago

Verified cd5e3b57

common : move gpt-oss reasoning processing to init params (#16937)

aldehir committed 55 days ago

Verified 87c9efc3

docs: remove llama_sampler_accept reference in sampling sample usage (#16920)

alundb committed 55 days ago

Verified 76af40aa

CUDA: add FLOOR, CEIL, ROUND, TRUNC unary ops (#16917)

mnehete32 committed 55 days ago

Verified 7db35a79

devops: fix failing s390x docker build (#16918)

taronaeo committed 55 days ago

Verified a864132b

ggml: add s390x cpu-feats (#16774)

taronaeo committed 55 days ago

Verified d38d9f08

scripts : add script to bench models (#16894)

ggerganov committed 55 days ago

Verified 7fd205a8

webui: auto-refresh /props on inference start to resync model metadata (#16784)

ServeurpersoCom committed 56 days ago

Verified 2f68ce7c

webui: add HTML/JS preview support to MarkdownContent with sandboxed iframe (#16757)

ServeurpersoCom committed 56 days ago

Verified e4a71599

vendor : update cpp-httplib to 0.27.0 (#16846)

angt committed 56 days ago

Verified dd5e8cab

Older