ngxson/llama.cpp

Pull Requests Commits

fix multi inheritance

ngxson committed 30 days ago

8fc86982

ngxson committed 30 days ago

53d149b5

Apply suggestions from code review

ngxson committed 30 days ago

Verified 3c9c8d9e

ngxson committed 31 days ago

4fd0ac71

Merge branch 'master' into xsn/qwen3a

ngxson committed 31 days ago

264371ca

Apply suggestions from code review

ngxson committed 31 days ago

Verified 46a8435d

mtmd: add Gemma 4 audio conformer encoder support (#21421)

stephencox-ict committed 31 days ago

Verified 547765a9

fix: Proper messages rendering for "Show raw output" (#21672)

allozaur committed 31 days ago

Verified 9e209c5a

docs: add guide on how to add multimodal support (#21778)

ngxson committed 31 days ago

Verified 6313acbe

CUDA: skip compilation of superfluous FA kernels (#21768)

JohannesGaessler committed 31 days ago

Verified ff5ef827

mtmd : add MERaLiON-2 multimodal audio support (#21756)

SiruiHe committed 32 days ago

Verified 073bb2c2

opencl: add basic support for q5_k (#21593)

shaofeiqi committed 32 days ago

Verified af1127d3

TP: fix Qwen 3 Next data split (#21732)

JohannesGaessler committed 32 days ago

Verified 865ff06b

ggml : fix a few instances of missing GGML_TYPE_Q1_0 cases (#21716)

CISC committed 32 days ago

Verified 2b2cd57d

py : Bump typer to latest to fix huggingface_hub issue (#21701)

bartowski1182 committed 32 days ago

Verified 660386f6

CUDA: also store node->src ne/nb for graph equality (#21736)

am17an committed 32 days ago

Verified a29e4c0b

fix: Fix broken structured output when using $refs in json_schema (#21699)

Galunid committed 32 days ago

Verified b136b62c

hexagon: add support for linux on snapdragon (#21707)

tboinovski1 committed 32 days ago

Verified 81069a80

hexagon: improved Op queuing, buffer and cache management (#21705)

max-krasnyansky committed 32 days ago

Verified 9aa28077

common : better align to the updated official gemma4 template (#21704)

aldehir committed 32 days ago

Verified 3fc65063

common : add callback interface for download progress (#21735)

angt committed 32 days ago

Verified 05b3caaa

model : make Gemma 4 shared-KV tail attn_k tensors optional on load (#21739)

MoonRide303 committed 32 days ago

Verified e62fa13c

ggml-webgpu: support non-square subgroup matrix configs for Intel GPUs (#21669)

SharmaRithik committed 32 days ago

Verified bfd1f453

ggml-webgpu: address quantization precision and backend lifecycle managment (#21521)

Constannnnnt committed 32 days ago

Verified e4fed9d0

server : ignore --alias when using --models-preset (#21380)

angt committed 32 days ago

Verified 5dd10253

common : fix when loading a cached HF models with unavailable API (#21670)

angt committed 32 days ago

Verified fb38d6f2

common: mark --split-mode tensor as experimental (#21684)

JohannesGaessler committed 33 days ago

Verified 0893f50f

webui: Static build output improvements (#21667)

allozaur committed 33 days ago

Verified f989a6e3

common : enable reasoning budget sampler for gemma4 (#21697)

berkidem committed 33 days ago

Verified d7ff074c

docs : fix broken link to ggml-openvino in OPENVINO.md (#21709)

ibelem committed 33 days ago

Verified 3f8752b5

Older