ggml-org/llama.cpp

Pull Requests Commits

ggerganov committed 34 days ago

Verified 121fe621

JohannesGaessler committed 34 days ago

803d3a19

JohannesGaessler committed 34 days ago

b90486e5

JohannesGaessler committed 34 days ago

7e466072

fix use-after-free in llama-model-loader.cpp

JohannesGaessler committed 34 days ago

4b7f407a

fixup for rebase

JohannesGaessler committed 34 days ago

e6a6af1c

tests: add end-to-end tests per model architecture

JohannesGaessler committed 34 days ago

fc696034

ggml-cpu: fix data race for debug asserts (#20148)

JohannesGaessler committed 34 days ago

Verified 2850bc6a

kv-cache : fix M-RoPE checkpoints (#20132)

ggerganov committed 34 days ago

Verified 17a42589

cli : Don't clear system prompt when using '/clear' (#20067)

roj234 committed 34 days ago

Verified f7db3f37

opencl: add neg, exp and diag (#20127)

lhez committed 34 days ago

Verified 6c97bffd

hexagon: add fp16 support for binary ops: add,sub,mul,div (#20139)

YardenTal44 committed 35 days ago

Verified 2b10b626

models : kda chunk size = 16 (#19827)

ymcki committed 35 days ago

Verified a0ed91a4

CUDA: Improve performance via less synchronizations between token (#17795)

aendk committed 35 days ago

Verified 2cd20b72

model : update Qwen3.5 model type detection (#20126)

EZForever committed 35 days ago

Verified 872646b3

cli : add command and file auto-completion (#19985)

CISC committed 35 days ago

Verified b5ed0e05

convert : register Qwen 3.5 ForCausalLM for text only (#20119)

CISC committed 35 days ago

Verified cf232515

webui: Improvements for Models Selector UI (#20066)

allozaur committed 35 days ago

Verified 5e335ba1

chore : correct typos [no ci] (#20041)

marcelpetrick committed 35 days ago

Verified 92f7da00

hexagon: Flash Attention optimizations (dma, mpyacc, multi-row) and MatMul updates (#20118)

max-krasnyansky committed 35 days ago

Verified 7a99dc85

opencl: add `SET`, support i32 for `CPY`, minor refactor for cpy (#20101)

lhez committed 35 days ago

Verified 69fd3453

hexagon: add llama-completion runner script (#20095)

tboinovski1 committed 36 days ago

Verified 1a29907d

[WebGPU] Fix wait logic for inflight jobs (#20096)

nikhilJain17 committed 36 days ago

Verified 24d2ee05

Add concat op to webgpu. (#20068)

yomaytk committed 36 days ago

Verified 541bf376

tools : add missing clocale include in mtmd-cli [no ci] (#20107)

CISC committed 36 days ago

Verified d969e933

ggml: fix ggml_is_contiguous_n for ne == 1 (#20092)

JohannesGaessler committed 36 days ago

Verified 7f5ee549

ggml : use a simple std::thread in AMX without OpenMP (#20074)

angt committed 36 days ago

Verified 66199c9f

impl : use 6 digits for tensor dims (#20094)

ddh0 committed 36 days ago

Verified c99909dd

Fix locale-dependent float printing in GGUF metadata (#17331)

ssam18 committed 36 days ago

Verified cb8f4fa3

completion : Fix a typo in warning message (#20082)

standby24x7 committed 36 days ago

Verified 54910bd4

Older