ggerganov/llama.cpp

Pull Requests Commits

jinja: use const for statement::execute and ::visit

ngxson committed 47 days ago

8c1536bc

convert: add MiMo-V2.6 support (#29257)

AesSedai committed 47 days ago

Verified bfd73a87

cmake : allow repeated find_package calls for llama (#29228)

miyanyan committed 47 days ago

Verified a60f9aea

chat : Fix Muse Glimmer tool-call first parser error (#29242)

NickM-27 committed 48 days ago

Verified 7ab4ee7b

ci : publish snapdragon builds in release workflow (#29007)

ykhrustalev committed 48 days ago

Verified 0ee9435b

Add close button to UI toasts (#28246)

agustinmista committed 48 days ago

Verified 8cfc315a

opencl: add A8 Q4_0 non-MoE dp4a binary kernel (#29055)

shaofeiqi committed 48 days ago

Verified ec5a12b8

ci : update Level Zero SDK to v1.33.1 and enable the L0/oneDNN CMake flags in the SYCL job (#29230)

Asahi-Prv committed 48 days ago

Verified c550d2f6

hexagon: new HMX-optimized GATED_DELTA_NET (#29199)

max-krasnyansky committed 48 days ago

Verified 58367713

vendor : update cpp-httplib to 0.57.1 (#29239)

angt committed 48 days ago

Verified ff0dbb97

metal : fix mask bounds in flash attention block pre-pass (#29220)

masterFoad committed 48 days ago

Verified fb34fc26

test-save-load-state : compare logits with NMSE and feed expected tokens (#29238)

ggerganov committed 48 days ago

Verified c641dfa8

llama-context : report graph inputs and input tensors during sched reserve (#26625)

ggerganov committed 48 days ago

Verified 96550613

cuda: fix sm_70 tile compilation error (#29224)

lingyezhixing committed 48 days ago

Verified b1c2863e

ggml-cuda : convert contiguous tensors four elements at a time (#29155)

pwilkin committed 48 days ago

Verified f4e276a2

cuda : accelerate conv2d with implicit GEMM (#29135)

leejet committed 48 days ago

Verified e6cef815

ggml : fix dimension and stride truncation in ggml_permute (#29227)

leejet committed 48 days ago

Verified c21284cd

vendor : update cpp-httplib to 0.57.0 (#29214)

angt committed 48 days ago

Verified 6f41ac59

docker : bump cuda to 13.4.1 (#29207)

CISC committed 48 days ago

Verified ec91ab5a

sycl : support gated DSV4_HC_PRE and optional HC_POST comb matrix (#29132)

cwriter committed 48 days ago

Verified bb3c853c

sycl : pinned memory use right device context instead of 0 (#28895)

lslusarczyk committed 48 days ago

Verified af911149

CUDA: Follow up of #25635, refactoring FA shared smem swizzle (#28536)

ynankani committed 48 days ago

Verified 1884824f

test-llama-archs : make tensor data stdev configurable and improve help (#29133)

ggerganov committed 48 days ago

Verified 161755f2

tests/test-backend-ops : allow regex entries in the -o filter (#29204)

ggerganov committed 48 days ago

Verified 1d72b05d

ci : refactor build-self-hosted into backend-specific workflows (#28991)

CISC committed 48 days ago

Verified 542e9202

args: add env vars for temperature, top-p, min-p and penalties (#27380)

kucharskim committed 48 days ago

Verified e0dff584

server : do not forward --api-key-file to router-spawned child instances (#28938)

nandan2003 committed 48 days ago

Verified 982a3329

tests : remove stale comment (#29140)

mostafafaheem committed 48 days ago

Verified 711f60be

ggml-metal : simplify fusion pattern op list declaration (#29206)

ggerganov committed 48 days ago

Verified 335b21fc

json: Fixed json enum handling (#28518)

Silverside committed 48 days ago

Verified 26394b4e

Older