ngxson/llama.cpp

Pull Requests Commits

ci : use actions from ggml-org

ngxson committed 341 days ago

f662391f

llama : avoid hardcoded QK_K (#11061)

ggerganov committed 341 days ago

Verified c07d437b

ggerganov committed 341 days ago

Verified 99a3755a

ggml : allow loading backend with env variable (ggml/1059)

rgerganov committed 341 days ago

Verified c792dcf4

ci : pin dependency to specific version (#11137)

ngxson committed 341 days ago

Verified 80ccf5d7

arg : option to exclude arguments from specific examples (#11136)

ggerganov committed 341 days ago

Verified a3c1232c

llamafile : ppc64le MMA INT8 implementation (#10912)

amritahs-ibm committed 341 days ago

Verified 8cef75c7

ci : fix cmake option (#11125)

ggerganov committed 341 days ago

Verified 0d52a69e

Disable GL_KHR_cooperative_matrix Vulkan extension if not available. (#11117)

mbaudier committed 341 days ago

Verified 02f04301

fix: Vulkan shader gen binary path when Cross-compiling (#11096)

ag2s20150909 committed 341 days ago

Verified bec2183f

GGUF: C++ refactor, backend support, misc fixes (#11030)

JohannesGaessler committed 342 days ago

Verified 53ff6b9b

ggml-backend : only offload from host buffers (fix) (#11124)

slaren committed 342 days ago

Verified 017cc5f4

ggml-backend : only offload from host buffers (#11120)

slaren committed 342 days ago

Verified a3d50bc0

rpc : code cleanup (#11107)

rgerganov committed 343 days ago

Verified a4dd4900

SYCL: Use get_multi_ptr instead of deprecated get_pointer in wkv6 (#11087)

qnixsynapse committed 343 days ago

Verified c0d6f790

llama-run : fix context size (#11094)

ericcurtin committed 343 days ago

Verified dc7cef9f

llama : remove unused headers (#11109)

ggerganov committed 343 days ago

Verified ecebbd29

github : add cmd line field to bug report (#11090)

ngxson committed 343 days ago

Verified 96be8c32

server : fix extra BOS in infill endpoint (#11106)

ggerganov committed 343 days ago

Verified e6e7c75d

llama : remove check flash_attn with lora (#11104)

ngxson committed 343 days ago

Verified 09186fab

llama : prevent system info string accumulation across calls (#11101)

a-ghorbani committed 343 days ago

Verified 96a1dc27

llama : rename missed batch params/vars to ubatch (#10059)

danbev committed 343 days ago

Verified 6369f867

llama : update llama_model API names (#11063)

ggerganov committed 343 days ago

Verified 47182dd0

tokenize : escape the prompt (#11058)

ggerganov committed 343 days ago

Verified 3e6e7a6b

mmap : fix fileno macro clash (#11076)

ggerganov committed 343 days ago

Verified ae2f606b

llama : use LLAMA_TOKEN_NULL (#11062)

ggerganov committed 343 days ago

Verified 727368c6

llama : use _impl suffix instead of _internal (#11060)

ggerganov committed 343 days ago

Verified 5047dd35

CUDA: add BF16 support (#11093)

JohannesGaessler committed 344 days ago

Verified 46e3556e

Vulkan: Add device-specific blacklist for coopmat for the AMD proprietary driver (#11074)

0cc4m committed 345 days ago

Verified b56f079e

llama : Add support for DeepSeek V3 (#11049)

fairydreaming committed 345 days ago

Verified 9394bbd4

Older