ggml-org/llama.cpp

Pull Requests Commits

ggerganov committed 247 days ago

Verified ed68474f

media : testing [no ci]

ggerganov committed 249 days ago

Verified a06d9a03

mtmd : Support Pixtral 12B (#13065)

ngxson committed 249 days ago

Verified ecda2ec4

convert : Append mult-eos,half-rope,bos to GLM4-0414 and Z (#13021)

piDack committed 249 days ago

Verified eb1776b1

rpc : add command line option for number of threads for the CPU backend (#13060)

rgerganov committed 249 days ago

Verified 2cca6c01

CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (#13014)

JohannesGaessler committed 250 days ago

Verified 658987cf

mtmd : support SmolVLM (version 1 and 2) (#13050)

ngxson committed 250 days ago

Verified dc39a5e7

security : add note about RPC and server functionality (#13061)

ggerganov committed 250 days ago

Verified ab47dec3

metal : add memory pool for temp allocs (#12850)

ggerganov committed 250 days ago

Verified 7b53389c

llava : update documentations (#13055)

ngxson committed 250 days ago

Verified 24345353

ggml : add SSE 4.2 and x64 base variant for CPUs without AVX (#12871)

slaren committed 251 days ago

Verified 1d735c0b

SYCL: Add non-contiguous support in ROPE (#12993)

qnixsynapse committed 251 days ago

Verified 5368ddda

mtmd : merge llava, gemma3 and minicpmv CLI into single `llama-mtmd-cli` (#13012)

ngxson committed 251 days ago

Verified 84a9bf2f

convert : experimental support for `--mmproj` flag (#13023)

ngxson committed 251 days ago

Verified 2016f07b

llava: fix errors in clip.h on certain compilers (#13030)

jmorganca committed 252 days ago

Verified 66023048

vulkan: support noncontiguous rms_norm (#13031)

jeffbolznv committed 252 days ago

Verified 66168204

metal: add neg operator (#13029)

jmorganca committed 252 days ago

Verified 4ba9d711

Disable CI cross-compile builds (#13022)

bandoti committed 253 days ago

Verified 00137157

gguf-py : fix upload python package workflow (#13020)

CISC committed 253 days ago

Verified fb28f4f8

clip : refactor, add `image_manipulation` and `llava_uhd` classes (#13011)

ngxson committed 253 days ago

Verified 37b9f0d2

main : Fix Ctrl+D/newline handling (#12951)

danielzgtg committed 253 days ago

Verified 64082100

gguf-py : GGUF Editor GUI - Python + Qt6 (#12930)

christopherthompson81 committed 254 days ago

Verified aff9d107

server : use std::move whenever possible (#12936)

ngxson committed 254 days ago

Verified 35370ba9

SYCL: Refactor and enable FP16 in binary broadcast OPs (#12975)

qnixsynapse committed 254 days ago

Verified 8d660057

mtmd : add methods to access `mtmd_image_tokens` (#12906)

ngxson committed 254 days ago

Verified b9154ecf

rpc : add RPC_CMD_HELLO (#12955)

rgerganov committed 254 days ago

Verified 2db9ba14

graph : make FA compatible with MLA + add initial Metal kernels (#12953)

ggerganov committed 255 days ago

Verified 2f74c354

ggml: Re-enable CUDA graphs in presence of CONT and DUP nodes (#12970)

agray3 committed 255 days ago

Verified 207c22ec

CANN: Add support for async operator submission (#12864)

hipudding committed 255 days ago

Verified 7a395f67

llama : recognize IBM Granite 3.3 FIM tokens (#12988)

Noeda committed 255 days ago

Verified 971f245b

Older