ggerganov/llama.cpp

Pull Requests Commits

vulkan: enable Conv2D for Apple after MoltenVK fixed the bug

0cc4m committed 74 days ago

7a152de3

model : add support for Seed-OSS (#15490)

pwilkin committed 74 days ago

Verified b1afcab8

scripts: fix compare-llama-bench.py (#15521)

JohannesGaessler committed 74 days ago

Verified 9ef53690

chat : fix debug build assertion in trim function (#15520)

LaffeyNyaa committed 74 days ago

Verified 21dc4dda

vulkan: Rewrite synchronization to allow some overlap between nodes (#15489)

jeffbolznv committed 74 days ago

Verified 289bf411

vulkan.Dockerfile: install vulkan SDK using tarball (#15282)

yeahdongcn committed 74 days ago

Verified b55f06e1

vulkan : support ggml_mean (#15393)

Acly committed 74 days ago

Verified 0a9b43e5

vulkan: optimize mul_mat_id loading row ids into shared memory (#15427)

jeffbolznv committed 74 days ago

Verified 330c3d2d

test-opt: allow slight inprecision (#15503)

JohannesGaessler committed 74 days ago

Verified e92734d5

ggml WebGPU: add support for quantization types (#15440)

reeselevine committed 74 days ago

Verified 45363632

model : gpt-oss add response_format support (#15494)

aldehir committed 75 days ago

Verified 32732f24

ggml: add `conv3d` op (#15182)

rmatif committed 75 days ago

Verified 92f7f0a5

cuda : add Pad Reflect 1D support (#14659)

YavorGIvanov committed 75 days ago

Verified b1ab9182

llama : remove KV cache defragmentation logic (#15473)

ggerganov committed 75 days ago

Verified 9ebebef6

ggml-cpu: Support Q5_0 and Q5_1 on s390x (#15486)

taronaeo committed 75 days ago

Verified ad5c975c

server : Support multimodal completion and embeddings prompts in JSON format (#15108)

65a committed 75 days ago

Verified 4afb0a74

readme : model : mtdm : lfm2 improvements (#15476)

tdakhran committed 75 days ago

Verified e2886936

CANN: Optimize RMS_NORM using cache (#15419)

noemotiovon committed 75 days ago

Verified a0f98dd6

sched : fix possible use of wrong ids tensor when offloading moe prompt processing (#15488)

slaren committed 75 days ago

Verified 54a241f5

llama : remove deprecated llama_kv_self API (#15472)

ggerganov committed 76 days ago

Verified cd36b5e5

graph : remove build_attn_with_sinks overload (#15469)

ggerganov committed 76 days ago

Verified 3f196be8

vulkan : support conv_2d_dw with f16 weights (#15392)

Acly committed 76 days ago

Verified 97ae5961

vulkan: add exp operation (#15456)

ddwkim committed 76 days ago

Verified 20c2dac8

vulkan: Reuse conversion results in prealloc_y (#15410)

jeffbolznv committed 76 days ago

Verified 96452a3f

examples : fix some typos in examples/model-conversion/README.md (#15477)

DamonFool committed 76 days ago

Verified 9ad5e60d

kv-cache : drop the "unified" prefix (#15467)

ggerganov committed 76 days ago

Verified 715a6db0

examples : install torch-cpu for model conversion tool/example (#15475)

DamonFool committed 76 days ago

Verified ad294df0

ci : enable RVV1.0 native build (#15386)

alitariq4589 committed 76 days ago

Verified 029bb39e

ci : continue file download with wget (#15471)

ggerganov committed 76 days ago

Verified 30649cab

examples : add model conversion tool/example (#15455)

danbev committed 76 days ago

Verified 2758fa10

Older