ggml-org/llama.cpp

Pull Requests Commits
Thinking model disabled assistant prefill (#15404)
gabe-l-hart committed 5 days ago
Verified 5fac79cb
Implement --log-colors with always/never/auto (#15792)
ericcurtin committed 5 days ago
Verified 408ff524
CUDA: fastdiv, launch bounds for mmvq + q8_1 quant (#15802)
JohannesGaessler committed 5 days ago
Verified 5143fa89
tests : add --list-ops and --show-coverage options (#15745)
danbev committed 5 days ago
Verified 3a550b5c
gguf: gguf_writer refactor (#15691)
Green-Sky committed 5 days ago
Verified a8128382
kv-cache : fix SWA checks + disable cacheless iSWA (#15811)
ggerganov committed 5 days ago
Verified c610b6c1
model-conversion : add --embeddings flag to modelcard.template [no ci] (#15801)
danbev committed 6 days ago
Verified 5d6688de
chat : fixed crash when Hermes 2 <tool_call> had a newline before it (#15639)
ExtReMLapin committed 6 days ago
Verified 4fd1242b
chat : nemotron thinking & toolcalling support (#15676)
pwilkin committed 6 days ago
Verified b2426e46
scripts : add Jinja tester PySide6 simple app (#15756)
pwilkin committed 6 days ago
Verified 9e2b1e83
llama : add support for EmbeddingGemma 300m (#15798)
danbev committed 6 days ago
Verified fb15d649
metal : Add template specialization for mul_mm_id w/ ne20 == 10 (#15799)
gabe-l-hart committed 6 days ago
Verified 856ed094
llama : set n_outputs to 1 to avoid 0 outputs mean-pooling (#15791)
danbev committed 6 days ago
Verified d1e2adba
CANN: Refactor ND to NZ workspace to be per-device (#15763)
noemotiovon committed 6 days ago
Verified c1c354e4
server: add exceed_context_size_error type (#15780)
ngxson committed 6 days ago
Verified a68d9144
Document the new max GPU layers default in help (#15771)
ericcurtin committed 6 days ago
Verified badb80ca
ggml: add ops for WAN video model (cuda && cpu) (#15669)
leejet committed 6 days ago
Verified 0a1b3982
CANN: Fix precision issue on 310I DUO multi-devices (#15784)
hipudding committed 6 days ago
Verified 5421f63a
opencl: add hs=40 to FA (#15758)
rmatif committed 6 days ago
Verified 820bc985
CANN: fix acl_rstd allocation size in ggml_cann_rms_norm (#15760)
noemotiovon committed 7 days ago
Verified 239b60e8
vulkan: fix mmv subgroup16 selection (#15775)
0cc4m committed 7 days ago
Verified dff7551b
vulkan: don't use std::string in load_shaders, to improve compile time (#15724)
jeffbolznv committed 7 days ago
Verified 0fce7a12
vulkan : update ggml_vk_instance_validation_ext_available (#15666)
danbev committed 7 days ago
Verified 8227695d
ggml vulkan: add hardsigmoid and hardswish operations (#15762)
relent95 committed 7 days ago
Verified 0014fb4a
CUDA: Optimize `rms_norm_f32` kernel and its fused variants, giving 1-6% perf E2E (#15715)
ORippler committed 7 days ago
Verified 661ae31c
model-conversion : fix pyright errors (#15770)
danbev committed 7 days ago
Verified 407c2378
sampling : optimize dist sampler (#15704)
ggerganov committed 7 days ago
Verified cdedb70a
replace M_SQRT2 with std::sqrt(2)
CISC committed 7 days ago
Verified 0408a4fa
Merge branch 'master' into cisc/grok-2
CISC committed 7 days ago
Verified c0d755cd
remove comment
CISC committed 7 days ago
Verified b7bfc9a6
Terms Of Use Privacy Policy Refund Policy Impressum