ggerganov/llama.cpp

Pull Requests Commits

use tensor split for deciding cache size

am17an committed 23 days ago

5b685032

ggerganov committed 23 days ago

Verified 0e7dac2d

cont : update arg help

ggerganov committed 23 days ago

Verified dbe8ef4b

am17an committed 23 days ago

ceb7b041

llama: support MoE cache over multiple GPUs

am17an committed 24 days ago

e67f8fa9

mtmd: add cohere2 vision support (#30062)

Terrencezzj committed 24 days ago

Verified 50a6c5cf

llama : add a GPU cache for MoE experts kept in host memory (#29887)

am17an committed 24 days ago

Verified d6cf9acb

cuda: update uncoalesced memory reads in pool2d (#29425)

shenron0101 committed 24 days ago

Verified 42c787e8

server : accumulate generated text and tokens as parse input (#29876)

aldehir committed 24 days ago

Verified 18b5f8b1

llama: share the nextn tensor flags between models (#30097)

ServeurpersoCom committed 24 days ago

Verified 448147d4

metal : few-row MMA mat-mul for the remaining src0 types (#30065)

pratiknarola-t committed 24 days ago

Verified 98819068

metal : fix MUL_MAT+ADD fusion when the residual is itself a MUL_MAT (#30100)

nyo16 committed 24 days ago

Verified 7e8324f5

feat: add GLM5Next MTP, optimize (#29928)

pwilkin committed 24 days ago

Verified b9acf138

qwen3tts : guard speaker_encoder_config patch for CustomVoice variant (#29179)

SIDDARTHAREDDY8 committed 24 days ago

Verified 48499d2e

sampling : use greedy selection for eligible temperature-zero chains (#29797)

hthadicherla committed 24 days ago

Verified d0b490f2

vocab : add plamo fim tokens (#30090)

CISC committed 24 days ago

Verified 42b021b4

convert : fix token configuration for PLaMo-3 (#29843)

tokinasin committed 24 days ago

Verified 7481354a

musa: use the tile lightning indexer kernel (#30080)

yeahdongcn committed 24 days ago

Verified ad215653

vendor : update cpp-httplib to 0.60.0 (#30081)

cabelo committed 24 days ago

Verified b7dafa01

imatrix : include clocale for std::setlocale (#30079)

dazzywi committed 24 days ago

Verified 26908739

ggml-webgpu: fix flash_attn supports_op check for overlapping KV (#28205)

yomaytk committed 24 days ago

Verified 36a73916

tests : retain the anchor when testing recurrent rollback (#29923)

eapache committed 24 days ago

Verified fa3c2fab

[SYCL] fix the issue in mixed different model GPUs in FA (#29071)

arthw committed 24 days ago

Verified 005a1e12

sycl: accelerate GLM MLA prefill with MKL flash attention (#29171)

anantshri committed 24 days ago

Verified d2a79e60

sycl : fattn_kv_buffers cleanup (#27689)

Pratyush-gg committed 24 days ago

Verified 5e5b628e

vulkan: fix amd iGPU slow checkpoint read (#30049)

0cc4m committed 24 days ago

Verified 4d756bc7

sycl: add IQ3_S multi-column MMVQ (#29500)

clemenswasser committed 24 days ago

Verified 78651c41

ggml-webgpu: no dawn native features on wasi (#27069)

MendyBerger committed 24 days ago

Verified f498f864

hexagon: CPY/CONCAT/CONT/DUP overhaul to use DMA/HVX for all cases (#30067)

max-krasnyansky committed 25 days ago

Verified c479922a

cuda : add BF16 support for XIELU (#29955)

Qiao12-pixel committed 25 days ago

Verified 5ad1c5da

Older