ggml-org/llama.cpp

Pull Requests Commits

explicit offline

ngxson committed 3 days ago

ff1566e2

ngxson committed 3 days ago

7680f3a4

restore timeout_seconds in some places

ngxson committed 3 days ago

a15f5c83

ngxson committed 3 days ago

3cd9b3de

server : speed up tests

ngxson committed 3 days ago

0be344dc

ggml-cpu: drop support for nnpa intrinsics (#15821)

taronaeo committed 4 days ago

Verified 186415d5

aLoRA Support (#15327)

gabe-l-hart committed 4 days ago

Verified fd621880

ci : exempt correct research label (#15825)

CISC committed 4 days ago

Verified 4281c7b3

Thinking model disabled assistant prefill (#15404)

gabe-l-hart committed 4 days ago

Verified 5fac79cb

Implement --log-colors with always/never/auto (#15792)

ericcurtin committed 4 days ago

Verified 408ff524

CUDA: fastdiv, launch bounds for mmvq + q8_1 quant (#15802)

JohannesGaessler committed 4 days ago

Verified 5143fa89

tests : add --list-ops and --show-coverage options (#15745)

danbev committed 4 days ago

Verified 3a550b5c

gguf: gguf_writer refactor (#15691)

Green-Sky committed 4 days ago

Verified a8128382

kv-cache : fix SWA checks + disable cacheless iSWA (#15811)

ggerganov committed 4 days ago

Verified c610b6c1

model-conversion : add --embeddings flag to modelcard.template [no ci] (#15801)

danbev committed 5 days ago

Verified 5d6688de

chat : fixed crash when Hermes 2 <tool_call> had a newline before it (#15639)

ExtReMLapin committed 5 days ago

Verified 4fd1242b

chat : nemotron thinking & toolcalling support (#15676)

pwilkin committed 5 days ago

Verified b2426e46

scripts : add Jinja tester PySide6 simple app (#15756)

pwilkin committed 5 days ago

Verified 9e2b1e83

llama : add support for EmbeddingGemma 300m (#15798)

danbev committed 5 days ago

Verified fb15d649

metal : Add template specialization for mul_mm_id w/ ne20 == 10 (#15799)

gabe-l-hart committed 5 days ago

Verified 856ed094

llama : set n_outputs to 1 to avoid 0 outputs mean-pooling (#15791)

danbev committed 5 days ago

Verified d1e2adba

CANN: Refactor ND to NZ workspace to be per-device (#15763)

noemotiovon committed 5 days ago

Verified c1c354e4

server: add exceed_context_size_error type (#15780)

ngxson committed 5 days ago

Verified a68d9144

Document the new max GPU layers default in help (#15771)

ericcurtin committed 5 days ago

Verified badb80ca

ggml: add ops for WAN video model (cuda && cpu) (#15669)

leejet committed 5 days ago

Verified 0a1b3982

CANN: Fix precision issue on 310I DUO multi-devices (#15784)

hipudding committed 5 days ago

Verified 5421f63a

opencl: add hs=40 to FA (#15758)

rmatif committed 5 days ago

Verified 820bc985

CANN: fix acl_rstd allocation size in ggml_cann_rms_norm (#15760)

noemotiovon committed 6 days ago

Verified 239b60e8

vulkan: fix mmv subgroup16 selection (#15775)

0cc4m committed 6 days ago

Verified dff7551b

vulkan: don't use std::string in load_shaders, to improve compile time (#15724)

jeffbolznv committed 6 days ago

Verified 0fce7a12

Older