ngxson/llama.cpp

Pull Requests Commits

chore: Update labeler to have separate labels for `server/webui` and `server` changes (#21567)

allozaur committed 59 days ago

Verified 3bd9aa1f

chore: Remove legacy files (#21606)

allozaur committed 60 days ago

Verified ece522f9

model : support step3-vl-10b (#21287)

forforever73 committed 60 days ago

Verified 09343c01

webui: fix syntax highlighting lost after streaming for non-common languages (#21206)

hmblair committed 60 days ago

Verified 97508acb

devops: kleidiai: provide KleidiAI-Enabled ARM Release Artifact (#21259)

martin-klacer-arm committed 60 days ago

Verified 5c4aae66

CUDA: make cuda graphs props check faster (#21472)

am17an committed 60 days ago

Verified c5ce4bc2

ggml-cuda: ds_read_b128 for q4_0 and q4_1 mmq kernels (#21168)

iacopPBK committed 60 days ago

Verified 66c4f9de

gguf-py : fix missing comma after bad merge in tensor-mapping (#21558)

danbev committed 60 days ago

Verified 93bdc615

kv-cache : support attention rotation for heterogeneous iSWA (#21513)

ggerganov committed 60 days ago

Verified 4eb19514

ggml-webgpu: parameterize submission size and add iOS specific limits (#21533)

reeselevine committed 60 days ago

Verified 957d717c

CUDA: check for buffer overlap before fusing (#21566)

am17an committed 60 days ago

Verified de1aa6fa

llama-server: fix model params not propagated (#21509)

taronaeo committed 60 days ago

Verified 69c28f15

unicode : add custom Qwen2 regex handler to fix segfault on long input (#21257)

nhs000 committed 60 days ago

Verified 0d049d6a

llama: remove per-arch tensor name lists (#21531)

JohannesGaessler committed 60 days ago

Verified a8ec0df4

server : fix restore for checkpoints with pos_min == 0 (#21510)

ggerganov committed 60 days ago

Verified e8f50826

ggml : deprecate GGML_OP_ADD1 (#21363)

ggerganov committed 60 days ago

Verified 22fc7913

ggml: Vulkan build, Linux -- output error string for errno on fork failure (#20868) (#20904)

tomoverlund committed 60 days ago

Verified 2a619f6f

vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl (#21029)

mkoker committed 60 days ago

Verified edd4d9bc

webui : store reasoning_content so it is sent back in subsequent requests (#21249)

aldehir committed 60 days ago

Verified 482192f1

ggml-cuda : fix CDNA2 compute capability constant for gfx90a (MI210) (#21519)

aviallon committed 60 days ago

Verified 71a81f6f

fix: Detect streaming state in reasoning content blocks (#21549)

allozaur committed 60 days ago

Verified ecce0087

Fix rtl text rendering (#21382)

Kabir08 committed 60 days ago

Verified d1f82e38

[SYCL] Add Q8_0 reorder optimization (~3x tg speedup on Intel Arc) (#21527)

PMZFX committed 61 days ago

Verified 0988accf

docs: fix typo in build.md (emdawbwebgpu -> emdawnwebgpu) (#21518)

CastelDazur committed 61 days ago

Verified 0033f53a

ggml-webgpu: Add the support of `MUL_MAT_ID` (#21147)

yomaytk committed 61 days ago

Verified d0a6dfeb

ggml: add Q1_0 1-bit quantization support (CPU) (#21273)

khosravipasha committed 61 days ago

Verified 2e1f0a88

cli: fix stripping of \n in multiline input (#21485)

bipinyadav3175 committed 61 days ago

Verified 506200cf

[CUDA ] Write an optimized flash_attn_stream_k_fixup kernel (#21159)

gaugarg-nv committed 61 days ago

Verified 15f786e6

llama-bench: add `-fitc` and `-fitt` to arguments (#21304)

am17an committed 61 days ago

Verified 94ca829b

vocab : add byte token handling to BPE detokenizer for Gemma4 (#21488)

aldehir committed 61 days ago

Verified 4aa962e2

Newer Older