ggml-org/llama.cpp

Pull Requests Commits

vulkan: tune MMVQ for Intel Windows

0cc4m committed 2 hours ago

98a04056

vendors : update miniaudio library to 0.11.24 (#19914)

data-man committed 3 hours ago

Verified 05728db1

vendor : update cpp-httplib to 0.35.0 (#19969)

angt committed 5 hours ago

Verified 4720819d

tests : model metadata loading from huggingface (#19796)

bartowski1182 committed 8 hours ago

Verified d979f2b1

CUDA: add CDNA3 MFMA support for flash attention MMA kernel (#19806)

Jayluci4 committed 23 hours ago

Verified ecbcb7ea

server: Add pragma once to server-context.h (#19944)

roj234 committed 1 day ago

Verified 3e6ab244

server: Mirroring /v1/responses to /responses to match /v1/chat/completions pattern (#19873)

samikama committed 1 day ago

Verified 5596a357

ci : use ubuntu-latest for gguf-publish workflow (#19951)

danbev committed 1 day ago

Verified 8d3b962f

ggml-cpu: add repack for mxfp4 (#19738)

am17an committed 1 day ago

Verified d903f30e

gguf-py : dump version to 0.18.0 (#19950)

danbev committed 1 day ago

Verified 8387ffb2

server : support multiple model aliases via comma-separated --alias (#19926)

ServeurpersoCom committed 1 day ago

Verified 2e7e6385

tests : enable test-chat out of tree build (#19558)

jplehr committed 1 day ago

Verified a8b192b6

replace the magic nunber 768 by max work group size to support iGPU (#19920)

arthw committed 1 day ago

Verified c17dce4f

ggml-zendnn: update code for latest ZenDNN API (#19923)

z-vishal committed 1 day ago

Verified 88cf781f

ggml : fix AMX and add batched support (#19925)

angt committed 1 day ago

Verified 4e76d24f

vulkan: fix fp16 Flash Attention on Windows AMD RDNA2 and below (#19921)

0cc4m committed 2 days ago

Verified 723c7106

mtmd : fix padding of n_tokens (#19930)

ggerganov committed 2 days ago

Verified 37964f44

server : fix ctx checkpoint restore logic (#19924)

ggerganov committed 2 days ago

Verified 01cd448b

kv-cache : fix can_shift() check to take into account M-RoPE (#19928)

ggerganov committed 2 days ago

Verified 99bd67c9

llama: Add option to merge gate and exp weights (#19139)

am17an committed 2 days ago

Verified b68d7516

ggml-virtgpu: improve the reliability of the code (#19846)

kpouget committed 2 days ago

Verified ffaafde1

server: fix load-on-startup not respected in ini file (#19897)

drrros committed 2 days ago

Verified efba35a8

jinja : correct default size for string slices (#19913)

EZForever committed 2 days ago

Verified 9b62913b

model : add Jina Embeddings v5 Nano (partial EuroBERT) support (#19826)

maximilianwerk committed 2 days ago

Verified 66287bda

gguf : avoid too many file size calls (#19919)

ggerganov committed 2 days ago

Verified 1ca3d1de

server : fix typo in server README.md (#19900)

yggdrasil75 committed 2 days ago

Verified bd723005

support permuted, remove check s0/s10 (#19889)

arthw committed 2 days ago

Verified 2943210c

vulkan: check for memory overlap before doing fusion (#19768)

jeffbolznv committed 3 days ago

Verified 3769fe6e

common : add more aliases for sampler CLI params (#19797)

ddh0 committed 3 days ago

Verified 832aa947

ci : update the ROCm/HIP toolchain versions [no ci] (#19891)

slojosic-amd committed 3 days ago

Verified 3af34b9f

Older