ngxson/llama.cpp

Pull Requests Commits

vulkan: dequantize iq4_xs 4 at a time (#20657)

netrunnereve committed 16 days ago

Verified 07feeaa9

cmake : fix build warning when kleidiai is enabled (#20457)

chaxu01 committed 16 days ago

Verified 3fee84e1

vocab : assert array size of scores and toktypes (#20737)

CISC committed 16 days ago

Verified 81139774

docs: add information about openvino in the docker page (#20743)

kannon92 committed 16 days ago

Verified c014c3f8

CANN: handle in-place ROPE on non-contiguous f32 tensors (#20274)

noemotiovon committed 16 days ago

Verified 7f2cbd9a

ggml-webgpu: Update the `RMS_NORM` preprocessor and add `L2_NORM` (#20665)

yomaytk committed 17 days ago

Verified 509a31d0

ggml-webgpu: Add supports for `DIAG` and `TRI` (#20664)

yomaytk committed 17 days ago

Verified ea01d196

CANN: support flash attention for head dim not multiple of 16, fix ALiBi slope offset (#20031)

noemotiovon committed 17 days ago

Verified 07ba6d27

model : add control vector support where missing (#20653)

GreyWorks committed 17 days ago

Verified 6729d492

gguf-py : cleaner way to get the first key (#20727)

CISC committed 17 days ago

Verified d13d60af

Rebuild index.html.gz (#20724)

crsawyer committed 17 days ago

Verified 5744d7ec

Move to no timeout for WaitAny in graph submission to avoid deadlocks in some cases on llvm-pipe backends (#20618)

reeselevine committed 17 days ago

Verified 8ced5f41

ggml-cpu/x86: fix unused changemask warning in repack (#20692)

mrshaw01 committed 17 days ago

Verified 78d550b5

ggerganov committed 17 days ago

4efd326e

ggml : bump version to 0.9.8 (ggml/1442)

ggerganov committed 17 days ago

b08f7322

ggml : restore ggml_type_sizef() to aboid major version bump (ggml/1441)

ggerganov committed 17 days ago

79187f2f

webui: improve tooltip wording for attachment requirements (#20688)

julien-c committed 17 days ago

Verified 48e61238

llama : re-enable manual LoRA adapter free (#19983)

PopFlamingo committed 17 days ago

Verified 312cf033

tests : fix test-jinja-py Windows failures by bypassing command-line args [no ci] (#20483)

rillomas committed 17 days ago

Verified f4049ad7

common : rework gpt-oss parser (#20393)

aldehir committed 17 days ago

Verified 5e8910a0

tests: enable kv_unified to prevent cuda oom error on rtx 2060 (#20645)

taronaeo committed 17 days ago

Verified fe00a84b

webui: Fix duplicated messages on q param (#20715)

allozaur committed 17 days ago

Verified 7ab321d4

HIP : ignore return of hipMemAdvise [no ci] (#20696)

IMbackK committed 17 days ago

Verified 7533a7d5

context : fix graph not resetting when control vector changes (#20381)

iimez committed 17 days ago

Verified a69d54f9

hexagon: add neg, exp, sigmoid, softplus ops, cont, repeat ops (#20701)

srikris-sridhar committed 18 days ago

Verified cf23ee24

vulkan: disable mmvq on Intel Windows driver (#20672)

0cc4m committed 18 days ago

Verified 892e3c33

ggml-blas: set mkl threads from thread context (#20602)

kannon92 committed 18 days ago

Verified ee4801e5

common/parser: add `--skip-chat-parsing` to force a pure content parser. (#20289)

pwilkin committed 18 days ago

Verified d2ecd2d1

ggml-cpu: fix RVV checks in quants and repacking (#20682)

taimur-10x committed 18 days ago

Verified 054d8b0f

ci : bump ccache [no ci] (#20679)

CISC committed 18 days ago

Verified ab0bb937

Newer Older