ggerganov/llama.cpp

Pull Requests Commits

register Qwen 3.5 ForCausalLM for text only

CISC committed 4 hours ago

Verified 4f1e02de

hexagon: add llama-completion runner script (#20095)

tboinovski1 committed 5 hours ago

Verified 1a29907d

[WebGPU] Fix wait logic for inflight jobs (#20096)

nikhilJain17 committed 8 hours ago

Verified 24d2ee05

Add concat op to webgpu. (#20068)

yomaytk committed 8 hours ago

Verified 541bf376

tools : add missing clocale include in mtmd-cli [no ci] (#20107)

CISC committed 14 hours ago

Verified d969e933

ggml: fix ggml_is_contiguous_n for ne == 1 (#20092)

JohannesGaessler committed 17 hours ago

Verified 7f5ee549

ggml : use a simple std::thread in AMX without OpenMP (#20074)

angt committed 17 hours ago

Verified 66199c9f

impl : use 6 digits for tensor dims (#20094)

ddh0 committed 19 hours ago

Verified c99909dd

Fix locale-dependent float printing in GGUF metadata (#17331)

ssam18 committed 19 hours ago

Verified cb8f4fa3

completion : Fix a typo in warning message (#20082)

standby24x7 committed 22 hours ago

Verified 54910bd4

docs: Fix intel documentation link (#20040)

mkdgs committed 1 day ago

Verified ecd99d6a

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

chaxu01 committed 1 day ago

Verified 137435ff

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

shaofeiqi committed 2 days ago

Verified 24350fdf

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

abhijitramesh committed 2 days ago

Verified 49a7564a

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

nikhilJain17 committed 2 days ago

Verified 4d828bd1

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

yomaytk committed 2 days ago

Verified 36a7a658

vulkan: tune MMVQ for Intel Windows (#19988)

0cc4m committed 2 days ago

Verified feefb928

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

angt committed 2 days ago

Verified ec88c3ce

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

taronaeo committed 2 days ago

Verified 2afcdb97

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

0cc4m committed 3 days ago

Verified 31914624

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

oobabooga committed 3 days ago

Verified 66d65ec2

vendors : update miniaudio library to 0.11.24 (#19914)

data-man committed 4 days ago

Verified 05728db1

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

angt committed 4 days ago

Verified 4720819d

tests : model metadata loading from huggingface (#19796)

bartowski1182 committed 4 days ago

Verified d979f2b1

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

Jayluci4 committed 5 days ago

Verified ecbcb7ea

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

roj234 committed 5 days ago

Verified 3e6ab244

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

samikama committed 5 days ago

Verified 5596a357

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

danbev committed 5 days ago

Verified 8d3b962f

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

am17an committed 5 days ago

Verified d903f30e

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

danbev committed 5 days ago

Verified 8387ffb2

Older