ggml-org/llama.cpp

Pull Requests Commits

Merge branch 'master' into xsn/mtmd_audio_gen_refactor

ngxson committed 1 day ago

f80943f3

server : fix usage.completion_tokens undercount with n > 1 (#29486)

unbug committed 1 day ago

Verified 9a3b91c6

server : add POST /tts endpoint for TTS models (#26603)

ngxson committed 1 day ago

Verified 5c0b49a2

server: add realtime transcription API (#30636)

ngxson committed 1 day ago

Verified 1242be66

refactor : move LFM2 shortconv and FFN blocks into a shared LFM base (#30874)

tdakhran committed 1 day ago

Verified 84e66d03

model : support Nemotron-3(.5)-Embed (#30877)

tokinasin committed 1 day ago

Verified 52ffa027

qwen: reshape output to 2d instead of 3d (#30823)

am17an committed 1 day ago

Verified 839a572a

mtmd: add state object for mtmd_gen

ngxson committed 1 day ago

c646c166

server: Support logprobs with speculative decoding (#27196)

pnb committed 1 day ago

Verified ead9f320

hexagon: fix VTCM size query on images that need layout out-params (#30655)

tdakhran committed 1 day ago

Verified b684b498

mtmd : lfm2 tiles and layout match transformers (#30261)

ykhrustalev committed 1 day ago

Verified 4a45de4f

model : add H2O-Lightning-4B decision model (#30361)

pseudotensor committed 1 day ago

Verified 7e3eeee0

mtmd: fix gemma 4 ffn act (#30882)

ngxson committed 1 day ago

Verified 5ba3b9fd

server : avoid repeated mmproj memory margin addition on sleep/resume (#28233)

TANGBUDU committed 1 day ago

Verified 5a9a34d8

ggml-cuda : support ssm_conv kernel size 7 (#30873)

tdakhran committed 1 day ago

Verified 39da130f

server: Router wasn't propagating the client disconnects fully to the child, the response wait could cause a delayed cancellation (#29707)

Me42019 committed 1 day ago

Verified 4e201388

server: fix missing output_index in Responses API SSE events for compatibility (#30863)

Renzishi committed 1 day ago

Verified 1ba568c5

mtmd : read LFM2 tiling params from GGUF metadata (#30840)

tdakhran committed 1 day ago

Verified 36140dbe

metal : optimize argmax and pool_1d (#30862)

ggerganov committed 1 day ago

Verified df117021

model : drop the redundant precision overrides in the Kolibri-1 router (#30865)

ggerganov committed 1 day ago

Verified 9df0b73c

ci : remove superfluous webgpu jobs (#30868)

CISC committed 1 day ago

Verified b38960f5

opencl: add several gemm variants for small batches and some fixes (#30850)

lhez committed 1 day ago

Verified a01ce9ff

sycl : replace memset due to not always ordered (#30812)

arthw committed 1 day ago

Verified 8115b4c3

ggml-cpu : select the NEON fp32->fp16 conversion (#30505)

bernardladenthin committed 1 day ago

Verified 119f68d7

HIP/CUDA: Fix reduction in mmf getting tile widths not devisiable by 64 (#30836)

IMbackK committed 1 day ago

Verified b6844985

ggml-zdnn: introduce flash attention op (#30512)

taronaeo committed 1 day ago

Verified 522f2b6e

spec : save the dynamic ngram cache file (#22055)

petersid2022 committed 1 day ago

Verified 83c2d012

CUDA: Clear CUDA graphs after legacy memory pool flush (#26771)

bokrosbalint committed 1 day ago

Verified 13786fbb

server : pass id_slot through the Anthropic /v1/messages conversion (#28554)

pcvelz committed 1 day ago

Verified d8061939

metal : fix missing MoE kernel variant for 9 experts per token (#30610) (#30699)

Caladrel65 committed 1 day ago

Verified 704366d7

Older