ngxson/llama.cpp

Pull Requests Commits

hip: compile debug builds with -O2 on hip to avoid a compiler bug (#20392)

IMbackK committed 27 days ago

Verified d63aa398

common/parser: add GigaChatV3/3.1 models support (#19931)

Mishusha committed 27 days ago

Verified a8304b4d

model : add support for Phi4ForCausalLMV (#20168)

dranger003 committed 27 days ago

Verified fdb17643

graph : add optional scale parameter to build_lora_mm [no ci] (#20427)

richarddd committed 27 days ago

Verified 1eea6a29

common : fix --n-cpu-moe, --cpu-moe for models with fused gate + up (#20416)

ddh0 committed 27 days ago

Verified 4a748b8f

ggml-webgpu: Add supports for `GGML_OP_REPEAT` (#20230)

yomaytk committed 27 days ago

Verified f2ab047f

llama : enable chunked fused GDN path (#20340)

ggerganov committed 27 days ago

Verified d28961d8

llama : whitespace cleanup (#20422)

CISC committed 27 days ago

Verified f90bd1dd

ggml : add NVFP4 quantization type support (#19769)

richarddd committed 27 days ago

Verified 5eae9cb1

benches : add nemotron super (#20420)

ggerganov committed 27 days ago

Verified 3ca19b0e

llama : add support for Nemotron 3 Super (#20411)

danbev committed 27 days ago

Verified eaf1d793

metal : fix capture_compute counter logic (#20410)

ggerganov committed 27 days ago

Verified 76ea1c1c

compare-llama-bench: check remotes as well (#20406)

am17an committed 27 days ago

Verified bd1ec818

metal : fix q5_k mul_mv register spill (#20399)

ggerganov committed 27 days ago

Verified b5412411

metal : add env var to trigger graph capture (#20398)

ggerganov committed 27 days ago

Verified c3632568

[SYCL] Update SYCL.md for binary package for Windows (#20401)

arthw committed 27 days ago

Verified ecac98ee

ci: disable coopmat on ubuntu-24-cmake-vulkan job (#20294)

0cc4m committed 27 days ago

Verified 182acfe5

common/parser: use nlohmann::ordered_json to preserve parameter order (#20385)

aldehir committed 27 days ago

Verified b5fe4559

common/parser: handle reasoning budget (#20297)

pwilkin committed 27 days ago

Verified acb7c790

ggml-cuda: gdn use shared mem for HIP (#20366)

IMbackK committed 28 days ago

Verified 5f91b1d5

cuda/hip: fix loop unrolling in ssm-conv (#20369)

IMbackK committed 28 days ago

Verified 9ef7523e

Fix agentic mcp image single model (#20339)

ServeurpersoCom committed 28 days ago

Verified 00de6153

vendor : update cpp-httplib to 0.37.0 (#20207)

cabelo committed 28 days ago

Verified e1a39999

vendor : update miniaudio to 0.11.25 (#20209)

cabelo committed 28 days ago

Verified 4f2f0a16

fix op rope, add rope_back (#20293)

arthw committed 28 days ago

Verified 0cec84f9

fix for failed UT case: ACC, L2_NORM, UPSCALE, fused_glu, unary (#20283)

arthw committed 28 days ago

Verified b2e1427c

model : qwen3vl reranker text support (#20332)

ViniciosLugli committed 28 days ago

Verified 4d99d450

llama-quant : correct `n_attention_wv` usage (#20357)

ddh0 committed 28 days ago

Verified 10e5b148

ggml : bump RPC version (#20330)

ggerganov committed 28 days ago

Verified 90b27318

ggml webgpu: faster normal quant and some k-quant matrix operations, better shader parameter handling (#20173)

reeselevine committed 28 days ago

Verified aa2d278a

Newer Older