ngxson/llama.cpp

Pull Requests Commits

change preprocess() call signature

ngxson committed 147 days ago

1915daea

rm unused clip_image_f32_batch::grid

ngxson committed 147 days ago

4b96581d

remove unused clip API

ngxson committed 147 days ago

706fba9f

ngxson committed 147 days ago

e981e5b7

add mtmd_image_preproc_out

ngxson committed 147 days ago

d1d735b6

mtmd: llava_uhd should no longer use batch dim (#24732)

ngxson committed 147 days ago

Verified f3e18281

ggml-cpu: Conditionally enable power11 backend based on compiler support (#24687)

shalinib-ibm committed 147 days ago

Verified 2e88c49c

metal : implement rope_back operator (#24725)

ggerganov committed 147 days ago

Verified 0843245c

metal : add f16 and bf16 support for concat operator (#24724)

ggerganov committed 147 days ago

Verified 8d2e5806

server: (router) add model management API (#23976)

ngxson committed 147 days ago

Verified 4b4d13ae

llama : skip main_gpu validation when no devices are available (#23405)

Dev-iL committed 147 days ago

Verified b4024af6

spec: fix segfault error on long prompts for eagle3 (#24707)

ruixiang63 committed 147 days ago

Verified 1a2dea29

[SYCL] add dev2dev memcpy by SYCL API (#24476)

arthw committed 147 days ago

Verified 74a80dd9

[SYCL] Add conv_3d (#24691)

arthw committed 147 days ago

Verified d1759e41

webui: export conversations as jsonl (#24688)

julien-c committed 147 days ago

Verified 8086439a

vulkan: record actual memory properties during buffer creation (#24326)

winstonma committed 147 days ago

Verified 558e221b

Revert "cuda: reset cuda context after reading memory size (#23935)" (#24715)

0cc4m committed 147 days ago

Verified ea21e039

ci: fix vulkan docker images (#24595)

Kononnable committed 147 days ago

Verified d5376cf5

UI : fix SSE transport detection and routing through CORS proxy. Assi… (#24500)

hrpnr committed 147 days ago

Verified bae36efa

opencl: optimize mul_mat_f16_f32_l4 for decode (#24504)

lhez committed 147 days ago

Verified 51571722

common: update logging to enforce max_capacity and optimize queue resizing (#24490)

max-krasnyansky committed 147 days ago

Verified cda63856

openvino: OV 2026.2, context-shift, Q5_1 support, gemma4 dense/embedding, and -fa off (#24503)

wine99 committed 147 days ago

Verified 890f1a27

sycl : Enable to support fp16 by OPs: SQR, SQRT, LOG, SIN, COS, CLAMP (#24692)

arthw committed 147 days ago

Verified 58728bdb

SYCL: fix use-after-free bug with async memcpy in MoE prefill (#24676)

sanmai committed 147 days ago

Verified ebbc1e51

sycl: Add optional USM system allocations (#22526)

ifdu committed 147 days ago

Verified 9b260fc9

vendor : update BoringSSL to 0.20260616.0 (#24693)

cabelo committed 148 days ago

Verified 74ade527

ui: add source toggle to mermaid and svg blocks (#24652)

ServeurpersoCom committed 148 days ago

Verified c1304d7b

Fix and restrict NVFP4 edge-cases in llama-graph (#24331)

ORippler committed 148 days ago

Verified 02810c7a

spec: add backend sampling support for eagle3 (#24655)

ruixiang63 committed 148 days ago

Verified a1824902

vulkan: prefer host-visible memory buffers on UMA devices (#22930)

winstonma committed 148 days ago

Verified 32120c10

Older