ggml-org/llama.cpp

Pull Requests Commits

docs: Fix intel documentation link (#20040)

mkdgs committed 91 days ago

Verified ecd99d6a

kleidiai : add sme fp16 compute path for q4_0 gemm on aarch64 (#20043)

chaxu01 committed 91 days ago

Verified 137435ff

opencl: add optimized q4_1 mm kernel for adreno (#19840)

shaofeiqi committed 91 days ago

Verified 24350fdf

ggml webgpu: fix workgroup dispatch limit for large batch sizes (#19965)

abhijitramesh committed 91 days ago

Verified 49a7564a

ggml webgpu: Clean up per-thread parameter buffer pool and job submission logic (#19772)

nikhilJain17 committed 92 days ago

Verified 4d828bd1

ggml-webgpu: Support non-contiguous `src0` and overlapping `src0/src1` in binary ops (#19850)

yomaytk committed 92 days ago

Verified 36a7a658

vulkan: tune MMVQ for Intel Windows (#19988)

0cc4m committed 92 days ago

Verified feefb928

scripts : improve get-wikitext-2.sh (#19952)

angt committed 92 days ago

Verified ec88c3ce

ggml-cpu: optimise s390x multiply extend instructions (#20032)

taronaeo committed 92 days ago

Verified 2afcdb97

vulkan: improve partial offloading performance on AMD (#19976)

0cc4m committed 93 days ago

Verified 31914624

cuda: cap grid.y at 65535 in non-contiguous dequantize/convert kernels (#19999)

oobabooga committed 93 days ago

Verified 66d65ec2

vendors : update miniaudio library to 0.11.24 (#19914)

data-man committed 94 days ago

Verified 05728db1

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

angt committed 94 days ago

Verified 4720819d

tests : model metadata loading from huggingface (#19796)

bartowski1182 committed 94 days ago

Verified d979f2b1

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

Jayluci4 committed 95 days ago

Verified ecbcb7ea

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

roj234 committed 95 days ago

Verified 3e6ab244

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

samikama committed 95 days ago

Verified 5596a357

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

danbev committed 95 days ago

Verified 8d3b962f

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

am17an committed 95 days ago

Verified d903f30e

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

danbev committed 95 days ago

Verified 8387ffb2

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

ServeurpersoCom committed 95 days ago

Verified 2e7e6385

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

jplehr committed 95 days ago

Verified a8b192b6

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

arthw committed 95 days ago

Verified c17dce4f

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

z-vishal committed 96 days ago

Verified 88cf781f

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

angt committed 96 days ago

Verified 4e76d24f

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

0cc4m committed 96 days ago

Verified 723c7106

mtmd : fix padding of n_tokens (#19930)

ggerganov committed 96 days ago

Verified 37964f44

server : fix ctx checkpoint restore logic (#19924)

ggerganov committed 96 days ago

Verified 01cd448b

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

ggerganov committed 96 days ago

Verified 99bd67c9

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

am17an committed 96 days ago

Verified b68d7516

Newer Older