ngxson/llama.cpp

Pull Requests Commits

ngxson committed 8 days ago

fc36eb77

add mtp graph for glm dense

ngxson committed 9 days ago

c256da1f

Merge branch 'master' into xsn/mtp_model

ngxson committed 9 days ago

6ded9d7b

vulkan: fix MMQ shader push constants and multi-dispatch (#19732)

0cc4m committed 9 days ago

Verified abb9f3c4

models : fix qwen3.5 beta/gate shapes (#19730)

ggerganov committed 9 days ago

Verified da348c9d

mtmd: build_attn modified, flash_attn on/off via ctx_params (#19729)

sfallah committed 9 days ago

Verified e6267a93

model : add JAIS-2 architecture support (#19488)

alielfilali01 committed 9 days ago

Verified 2bf318fd

CUDA: fix kernel selection logic for tile FA (#19686)

JohannesGaessler committed 9 days ago

Verified c78e6822

mtmd : chat : Fix extra \n between text and media marker (#19595)

tdakhran committed 9 days ago

Verified c5897995

webui: Fix Attachments not being included in completion request (#19731)

allozaur committed 9 days ago

Verified 03fd9d3b

model : add tokenizer from LFM2.5-Audio-1.5B (#19687)

tdakhran committed 9 days ago

Verified 8004f3a8

llama : use output_resolve_row() in get_logits_ith/get_embeddings_ith (#19663)

danbev committed 9 days ago

Verified eacb4b67

model : full modern bert support (#18330)

ryan-mangeno committed 9 days ago

Verified c0d04303

llamafile: powerpc: add FP16 MMA path for Q4/Q8 matmul (#19709)

shalinib-ibm committed 9 days ago

Verified 3bb2fcc8

models : dedup qwen35 graphs (#19660)

ggerganov committed 9 days ago

Verified 27326bfc

models : dedup Kimi Linear delta net implementation (#19668)

ymcki committed 9 days ago

Verified ad9f692f

Add Jinja support for "indent" string filter (#19529)

pwilkin committed 9 days ago

Verified 8a709735

ggml webgpu: Fix bug in dispatching large matrix-vector multiplication (#19535)

reeselevine committed 9 days ago

Verified e7f2f95c

server: save generated text for the /slots endpoint (for LLAMA_SERVER_SLOTS_DEBUG=1) (#19622)

matteoserva committed 10 days ago

Verified b55dcdef

model: support GLM-OCR (#19677)

ngxson committed 10 days ago

Verified eeef3cfc

docs: Fix broken links for preparing models in Backends (#19684)

MaciejDromin committed 10 days ago

Verified e99f1083

ggml webgpu: shader library organization (#19530)

reeselevine committed 10 days ago

Verified 238856ec

Pre-MCP UI and architecture cleanup (#19689)

allozaur committed 10 days ago

Verified ea003229

vulkan: split mul_mat into multiple dispatches to avoid overflow (#19509)

jeffbolznv committed 10 days ago

Verified d0061be8

common : make small string helpers as inline functions (#19693)

angt committed 10 days ago

Verified a569bda4

opencl: refactor expm1 and softplus (#19404)

shaofeiqi committed 10 days ago

Verified e2f19b32

opencl: optimize mean and sum_row kernels (#19614)

shaofeiqi committed 10 days ago

Verified 983559d2

model-conversion : add option to print tensor values (#19692)

danbev committed 10 days ago

Verified 2b089c77

Pre-MCP UI and architecture cleanup (#19685)

allozaur committed 11 days ago

Verified afa6bfe4

ggml: ggml-cpu: force-no-lto-for-cpu-feats (#19609)

talhaHavadar committed 11 days ago

Verified ae2d3f28

Older