ngxson/llama.cpp

Pull Requests Commits

ngxson committed 9 days ago

0a63bd80

Merge branch 'master' into xsn/server_improve_spec

ngxson committed 9 days ago

afe25301

cuda : add FILL op support (#17851)

JayZenith committed 9 days ago

Verified 51e0c2d9

server : add development documentation (#17760)

ngxson committed 9 days ago

Verified 37a4f632

server : make cache_reuse configurable per request (#17858)

ggerganov committed 9 days ago

Verified 2bc96931

Merge branch 'master' into xsn/server_improve_spec

ngxson committed 9 days ago

ba5c0b42

cuda: optimize SOLVE_TRI using registers and FMAF (#17703)

wsbagnsv1 committed 9 days ago

Verified 5814b4dc

ggml-cpu: add ggml_thread_cpu_relax with Zihintpause support (#17784)

ixgbe committed 9 days ago

Verified 79d61896

server : add comment

ggerganov committed 9 days ago

Verified 75be6ba0

server : fix draft acceptance logs (add SRV_CNT, SLT_CNT macros)

ggerganov committed 9 days ago

Verified f74d1ee9

server : fix generation time measurement

ggerganov committed 9 days ago

Verified 084cec95

model: add llama 4 scaling for mistral-large (deepseek arch) (#17744)

ngxson committed 10 days ago

Verified 4d372627

Vulkan: improve mul_mat_vec_iq1_m (#16907)

lovedheart committed 10 days ago

Verified 08f9d3cc

ci : add windows-cuda 13.1 release (#17839)

CISC committed 10 days ago

Verified 0a540f9a

common : change --color to accept on/off/auto, default to auto (#17827)

CISC committed 11 days ago

Verified 22577583

sycl: add missing BF16 conversion support for Intel oneAPI (#17780)

yingying0906 committed 11 days ago

Verified d9e03db1

vulkan: perf_logger improvements (#17672)

jeffbolznv committed 11 days ago

Verified db978373

ggml-zendnn : add ZenDNN backend for AMD CPUs (#17690)

z-vishal committed 11 days ago

Verified 017761da

add link to the PR

ngxson committed 11 days ago

398ae8db

server: support multiple generations from one prompt (OAI "n" option) (#17775)

ngxson committed 11 days ago

Verified c42712b0

fix small draft case

ngxson committed 11 days ago

cac8d7b2

ggml : add circular tiling support to pad, for Vulkan, CUDA, and CPU (used for making seamless textures) (#16985)

Phylliida committed 11 days ago

Verified 09c7c50e

HIP: fix RDNA3 FP16/BF16 matrix multiplication (#17817)

JohannesGaessler committed 11 days ago

Verified f334b794

webui: Stop generation from chat sidebar (#17806)

allozaur committed 11 days ago

Verified a28e3c75

webui: Fix context available value in Multi-model Router mode (#17804)

allozaur committed 11 days ago

Verified e31b5c55

webui: Per-conversation system message with UI displaying, edition & branching (#17275)

allozaur committed 11 days ago

Verified 21f24f27

ggml : improve error handling for search path existence checks (#17653)

flyinskyin2013 committed 11 days ago

Verified 7b43f557

llama : remove quantization sanity check (#17788)

danbev committed 11 days ago

Verified 444f00b0

vulkan: Use one row per workgroup for f32 mmv (#17711)

jeffbolznv committed 11 days ago

Verified 2960eb29

convert: support Mistral 3 Large MoE (#17730)

ngxson committed 11 days ago

Verified dbc15a79

Older