ngxson/llama.cpp

Pull Requests Commits

convert: add error message for mistral3 quantized weight

ngxson committed 68 days ago

0d7c1435

server: explicitly set exec path when create new instance (#17669)

ngxson committed 68 days ago

Verified 682e6658

ci : skip winget update when not in ggml-org (#17465)

angt committed 68 days ago

Verified 4574f294

ggml : add fallback definition for HWCAP2_SVE2 (#17683)

angt committed 68 days ago

Verified ab6726ee

Add context info to server error (#17663)

allozaur committed 68 days ago

Verified cee92af5

ggml-cuda: reorder only relevant nodes (#17639)

am17an committed 69 days ago

Verified ed320899

release: fix duplicate libs, store symbolic links (#17299)

taronaeo committed 69 days ago

Verified 7b6d7453

enhance argsort for UT (#17573)

NeoZhangJianyu committed 69 days ago

Verified 98bd9ab1

Override SSM_A op for Qwen3 Next to reduce splits (#17587)

pwilkin committed 69 days ago

Verified 746f9ee8

ops.md: update vulkan support (#17661)

jeffbolznv committed 69 days ago

Verified 9810cb82

mtmd: add mtmd_context_params::warmup option (#17652)

ngxson committed 69 days ago

Verified ecf74a84

fix: llama arch implementation (#17665)

giladgd committed 69 days ago

Verified 00c361fe

server: introduce API for serving / loading / unloading multiple models (#17470)

ngxson committed 69 days ago

Verified ec18edfc

common: improve verbosity level definitions (#17630)

ngxson committed 69 days ago

Verified 77334097

model: support Ministral3 (#17644)

ngxson committed 69 days ago

Verified cd3c1189

metal : add FA head size 48 (#17619)

ggerganov committed 69 days ago

Verified 649495c9

ggml : extend the GGML_SCHED_NO_REALLOC debug logic of the scheduler (#17617)

ggerganov committed 69 days ago

Verified 90c72a61

llama-graph: avoid expand_forward for fusion (#17633)

am17an committed 69 days ago

Verified 6eea6669

contributing: update guidelines for AI-generated code (#17625)

ngxson committed 70 days ago

Verified ff90508d

cmake : add option to build and link LibreSSL (#17552)

angt committed 70 days ago

Verified 0a4aeb92

model: LFM2-VL fixes (#17577)

tdakhran committed 70 days ago

Verified 2ba71951

clip: fix nb calculation for qwen3-vl (#17594)

ngxson committed 70 days ago

Verified 7f8ef50c

cli: add migration warning (#17620)

ngxson committed 70 days ago

Verified 3c136b21

common : throttle download progress output to reduce IO flush (#17427)

angt committed 70 days ago

Verified beb1f0c5

common: add LLAMA_LOG_FILE env var (#17609)

taronaeo committed 70 days ago

Verified def5404f

ggml: fix: macOS build with `-DGGML_BACKEND_DL=ON` (#17581)

giladgd committed 71 days ago

Verified fa046595

common: update env var name (#17588)

ddh0 committed 71 days ago

Verified 5a6241fe

CUDA: add stream-based concurrency (#16991)

am17an committed 71 days ago

Verified c7af376c

cuda : add error checking for cudaMemcpyAsync in argsort (#17599)

Mahekk357 committed 71 days ago

Verified 00425e2e

vulkan : fix FA mask load with bounds check (coopmat2) (#17606)

Acly committed 71 days ago

Verified 385c3da5

Older