ngxson/llama.cpp

Pull Requests Commits

ngxson committed 82 days ago

edc53a99

Merge pull request #31 from rujialiu/mrope-fix

ngxson committed 82 days ago

Verified ef1af68f

embedding: add raw option for --embd-output-format (#16541)

SamMalayek committed 82 days ago

Verified 1c1409e1

llama: consistent ctx <-> buf order for KV cache (#16746)

JohannesGaessler committed 82 days ago

Verified 7a0e900e

grammar : support array references in json schema (#16792)

aldehir committed 82 days ago

Verified 280d97be

CANN: Improve device ID handling and aclnnArange checks (#16752)

noemotiovon committed 82 days ago

Verified 3479efd1

CUDA: add unused vars to mmvf and mmvq (#16807)

am17an committed 82 days ago

Verified 463bbf20

sycl: add SSM_CONV operation support (#16800)

tamarPal committed 82 days ago

Verified ad8d36be

chat: Add LFM2 tool handling (#16763)

ykhrustalev committed 82 days ago

Verified c053e18a

mtmd : fix idefics3 preprocessing (#16806)

ngxson committed 82 days ago

Verified e1ab0848

llama : disable pipeline parallelism if compute buffer allocation fails (#16748)

slaren committed 83 days ago

Verified 5a4ff43e

ggml : fix interpolate with align-corners and ne=1 (#16700)

Acly committed 83 days ago

Verified 10640e31

HIP: fix AMDGPU_TARGETS, update documentation (#16803)

JohannesGaessler committed 83 days ago

Verified 80d28f10

model : add LightOnOCR-1B model (#16764)

ngxson committed 83 days ago

Verified c55d53ac

llama: fix leaked buffers for mmap + split files (#16765)

JohannesGaessler committed 83 days ago

Verified 945501f5

test-backend-ops: print failed tests at the end (#16785)

am17an committed 83 days ago

Verified 75cbdd3f

sycl: add ROLL operation support (#16665)

tamarPal committed 83 days ago

Verified 2b9bd9bf

sycl: add REPEAT_BACK operation support (#16734)

shani-f committed 83 days ago

Verified 59fc1ec8

CUDA: support for weight clamp in top-k norm (#16702)

am17an committed 83 days ago

Verified 75d33b93

ggml-alloc : make gallocr prefer chunks that allow memory reuse (#16788)

Acly committed 83 days ago

Verified 3470a5c8

cuda : use fast copy when src and dst are of different type and contiguous (#16789)

CISC committed 84 days ago

Verified bd562fe4

ggml: fix cuda kernel launch configuration for k_compute_batched_ptrs to support large batch (#16744)

leejet committed 84 days ago

Verified bbac6a26

convert : enable expert group selection for all models with it (#16691)

CISC committed 84 days ago

Verified 73a48c97

graph : add clamping to ffn_moe_weights_sum to avoid div-by-zero (#16655)

CISC committed 84 days ago

Verified f696428c

model : set res->t_embd in SmallThinker models (#16782)

CISC committed 84 days ago

Verified 7cce4f81

docs : add Jamba to Text-only models list (#16778)

amirai21 committed 84 days ago

Verified 8d886282

CUDA: General GEMV fusion (#16715)

am17an committed 84 days ago

Verified f77c13b9

vulkan: deduplicate Microsoft Direct3D12 devices (#16689)

giladgd committed 84 days ago

Verified 3cfa9c3f

convert : handle mmproj filename/path properly (#16760)

Galunid committed 85 days ago

Verified 5d195f17

model : set res->t_embd in PLaMo2 models (#16766)

mitmul committed 85 days ago

Verified 226f295f

Older