ggerganov/llama.cpp

Pull Requests Commits

metal : use F16 math in mul_mat kernels

ggerganov committed 344 days ago

Verified 4ff0831c

metal : minor code formatting

ggerganov committed 344 days ago

Verified b7564411

[SYCL] Fix building Win package for oneAPI 2025.0 update (#10483)

NeoZhangJianyu committed 344 days ago

Verified 5a898779

speculative : refactor and add a simpler example (#10362)

ggerganov committed 344 days ago

Verified d9d54e49

flake.lock: Update (#10470)

ggerganov committed 345 days ago

Verified cce5a900

llama : fix op mul check with command-r-plus (#10476)

slaren committed 345 days ago

Verified dc39012c

convert : XLMRoberta Type Vocab Size (#10458)

gabe-l-hart committed 345 days ago

Verified 9336db46

fix gguf-py: Conversion error when multiple licenses are configured (#9807)

mmngays committed 345 days ago

Verified 96fa2c5e

ggml : do not use ARM features not included in the build (#10457)

slaren committed 346 days ago

Verified 55ed008b

ci: Update oneAPI runtime dll packaging (#10428)

MakeDecisionWorth committed 347 days ago

Verified 6dfcfef0

GitHub: ask for more info in issue templates (#10426)

JohannesGaessler committed 347 days ago

Verified 599b3e0c

CANN: Support Ascend310P to accelerate F32 and F16 Model (#10216)

leo-pony committed 347 days ago

Verified c18610b4

cuda : optimize argmax (#10441)

slaren committed 348 days ago

Verified a5e47592

llama : handle KV shift for recurrent models (#10402)

ggerganov committed 348 days ago

Verified 1bb30bf2

ggerganov committed 348 days ago

Verified 87a533be

ggml/sched : do not skip views in pre-assignments

slaren committed 348 days ago

Verified 59b91728

ggml-opt: fix data corruption (ggml/1022)

JohannesGaessler committed 348 days ago

Verified 02e4eaf2

vulkan: predicate max operation in soft_max shaders/soft_max (#10437)

jeffbolznv committed 349 days ago

Verified 9abe9eea

cmake: add link dependencies to cmake find pkg (#10433)

bandoti committed 349 days ago

Verified f95caa79

llama : add .clang-format file (#10415)

slaren committed 349 days ago

Verified fab5d30f

vulkan: copy iq4_nl LUT into shared memory (#10409)

jeffbolznv committed 349 days ago

Verified 8fd4b7fa

vulkan: further optimize mul_mat_vec using larger loads (#10387)

jeffbolznv committed 349 days ago

Verified 1bacb9f6

update rel to 4040 (#10395)

NeoZhangJianyu committed 349 days ago

Verified ad21c9e1

Fix missing file renames in Makefile due to changes in commit ae8de6d50a (#10413)

avdg committed 349 days ago

Verified 3952a221

add cmake rvv support (#10411)

lhpqaq committed 350 days ago

Verified 42ae10bb

ggerganov committed 350 days ago

9fe0fb06

metal : fox offset integer overflows in im2col (ggml/1015)

pminev committed 350 days ago

611fabd7

metal : add `GGML_UNARY_OP_ELU` kernel (ggml/1018)

PABannier committed 350 days ago

12b0ad95

cmake: force MSVC compiler charset to utf-8 (#9989)

MakeDecisionWorth committed 350 days ago

Verified 342397dc

Add required ggml-base and backend libs to cmake pkg (#10407)

bandoti committed 350 days ago

Verified 2a11b6b0

Older