ggml-org/llama.cpp

Pull Requests Commits

add windows behavior check

CISC committed 6 hours ago

Verified ccb3f064

mtmd: add mtmd_helper_get_capabilities (#31183)

ngxson committed 7 hours ago

Verified c6fb3b5b

ggml-cpu : fix unused variables in arm arch (#31180)

CISC committed 8 hours ago

Verified 8b28c239

tests : refactor test-llama-archs to not link internal libllama symbols (#30880)

ggerganov committed 8 hours ago

Verified fd8f8451

ci : fix race in webgpu memset (#31162)

eapache committed 9 hours ago

Verified 9125a2eb

metal : stop overflow from happening in large matrix output offsets (#30072)

feni6 committed 10 hours ago

Verified 9b8446b9

ui: UX improvements (#31030)

allozaur committed 10 hours ago

Verified dc9e3bed

ui: Sidebar resizing improvements (#31119)

allozaur committed 10 hours ago

Verified 69b3cc40

ci : bump server timeout during startup (#31174)

ggerganov committed 10 hours ago

Verified 224d94c0

ci : drop the `macos-latest-x64` job (#31175)

ggerganov committed 10 hours ago

Verified 48c36110

deepseekv4.1: follow-up review fixes (#31137)

am17an committed 11 hours ago

Verified a0742153

cuda : use std::string instead of VLA for archName (#30463)

angt committed 19 hours ago

Verified d854d98b

common, grammar: add recursive depth limit, make it a requirement (#31153)

ngxson committed 20 hours ago

Verified 48c9f08b

tests : adjust default threads for test-llama-context (#31157)

ggerganov committed 1 day ago

Verified 6050b467

qwen: merge tensors on convert (#31028)

am17an committed 1 day ago

Verified 2280f0fd

ci : disable test-llama-context with fa off in the webgpu (#30977)

yomaytk committed 1 day ago

Verified 1a0b56ce

llama: fix exhausting CPU and MEM caused by dry sequence breaker (#31035)

trunk-trick committed 1 day ago

Verified 63412d87

llama-quant : fix 4D tensor quantization using ggml_nrows() and name-based expert detection (#30820)

anthony-wss committed 1 day ago

Verified 5c965fd8

grammar: limit recursion depth to prevent stack overflow (#29495) (#29755)

Hrushikesh-ramilla committed 1 day ago

Verified f1168ac4

llama : respect no_alloc in all memory caches (#31087)

ggerganov committed 1 day ago

Verified fa3645fe

tools : reject truncated GGUF tensor data in merge (#31044)

shreya-11733 committed 1 day ago

Verified 7790a923

CUDA: fuse FFN gate/up matmuls and GLU in MMQ with interleaved gate/up rows (#29948)

anujj committed 1 day ago

Verified c49edf09

ggml-cuda: fuse qwen MRoPE + set-rows (#31025)

am17an committed 1 day ago

Verified a55c16e4

vulkan: Add a larger tile size for conv3d (coopmat2 only) (#29920)

jeffbolznv committed 1 day ago

Verified e321bdad

vulkan : one warp tile per subgroup in the small matmul tiles (#30106)

ianloic committed 1 day ago

Verified 459a1ba5

hexagon: generic dims collapse/reshape for unary, binary and matmul ops (#31071)

max-krasnyansky committed 1 day ago

Verified f0fb5984

opencl: add topk and lightning indexer (#31026)

lhez committed 1 day ago

Verified 9a4580b6

opencl: add initial q2_k support (#30999)

lhez committed 1 day ago

Verified b083c384

ci : update CPU job assignments (#31136)

ggerganov committed 1 day ago

Verified 344be0a6

chat : reject JSON nested more than 128 levels before walking it (#30242)

ykhrustalev committed 1 day ago

Verified b3fa6995

Older