ngxson/llama.cpp

Pull Requests Commits

wip deepseek ocr

ngxson committed 78 days ago

554a46db

ggml-alloc : fix leak when reusing a tensor with a larger size (#16679)

slaren committed 79 days ago

Verified b617cfd2

Prevent premature submission on IME input (#16673)

allozaur committed 79 days ago

Verified 79068501

Import/Export UX improvements (#16619)

allozaur committed 79 days ago

Verified 0e4a0cf2

Enable per-conversation loading states to allow having parallel conversations (#16327)

allozaur committed 79 days ago

Verified 13f2cfad

llama-batch: fix build fails with `-Werror=missing-braces` (#16614)

otegami committed 79 days ago

Verified 06332e28

readme: update bindings (#16651)

deadprogram committed 79 days ago

Verified 72d53e6c

SYCL: Add support for FLOOR,CEIL,ROUND and TRUNC unary operators (#16613)

safranowith committed 79 days ago

Verified 2330de7b

llama-context: only warn on pooling_type when user specified (#16674)

otegami committed 79 days ago

Verified 7062dd84

model : add Granite Hybrid types (#16635)

giuseppe committed 80 days ago

Verified 0398752d

ci : fix binaries release failure for s390x (binaries may not work yet) (#16664)

taronaeo committed 80 days ago

Verified 4f73d0a9

ci : avoid manual updates of docs/ops.md (#16663)

CISC committed 80 days ago

Verified cec5edbc

ci: include s390x release binaries (#16648)

taronaeo committed 80 days ago

Verified fcb235b4

CODEOWNERS: update for ggml-cuda/mmf (#16660)

am17an committed 80 days ago

Verified 55754beb

HIP: fix GPU_TARGETS (#16642)

JohannesGaessler committed 81 days ago

Verified ee09828c

vulkan: Implement topk_moe fused shader, ported from CUDA (#16641)

jeffbolznv committed 81 days ago

Verified e56abd20

CUDA: use registers instead of smem in topk-moe (#16647)

am17an committed 81 days ago

Verified 38355c6c

opencl: transposed gemm/gemv moe kernel with mxfp4,f32 (#16602)

shawngu-quic committed 81 days ago

Verified 81387858

llama-model: fix insonsistent ctxs <-> bufs order (#16581)

JohannesGaessler committed 82 days ago

Verified 66b0dbcb

rpc : report actual free memory (#16616)

rgerganov committed 82 days ago

Verified 41386cf3

vulkan: Add State Space Model (SSM) Operations Support (#16463)

giuseppe committed 82 days ago

Verified 3d4e86bb

ggml : fix SpaceMit IME array out-of-bounds in task assignment (#16629)

muggle-stack committed 82 days ago

Verified 342c728d

webui: reorganize settings layout (#16607)

ServeurpersoCom committed 82 days ago

Verified ababae7e

vulkan: fix debug build (add_rms_len/data not found) (#16624)

jeffbolznv committed 82 days ago

Verified b1949159

metal : add `CONV_TRANSPOSE_2D` (#16542)

iliailmer committed 82 days ago

Verified 9ad4f193

grammar : use int64_t to avoid int overflows in int schema to grammar conversion logic (#16626)

ochafik committed 82 days ago

Verified 79967ec5

SYCL SET operator optimized for F32 tensors (#16350)

GittyBurstein committed 82 days ago

Verified ceff6bb2

mtmd : support home-cooked Mistral Small Omni (#14928)

ngxson committed 83 days ago

Verified 1bb4f433

fix: added a normalization step for MathJax-style \[\] and \(\) delimiters (#16599)

ServeurpersoCom committed 83 days ago

Verified 683fa6ba

sycl : add ARANGE operator (#16362)

GittyBurstein committed 83 days ago

Verified b22572e9

Older