ggerganov/llama.cpp

Pull Requests Commits

metal : fix capture_compute counter logic

ggerganov committed 10 days ago

Verified 8c8544f9

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

am17an committed 10 days ago

Verified bd1ec818

metal : fix q5_k mul_mv register spill (#20399)

ggerganov committed 10 days ago

Verified b5412411

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

ggerganov committed 10 days ago

Verified c3632568

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

arthw committed 10 days ago

Verified ecac98ee

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

0cc4m committed 10 days ago

Verified 182acfe5

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

aldehir committed 10 days ago

Verified b5fe4559

common/parser: handle reasoning budget (#20297)

pwilkin committed 10 days ago

Verified acb7c790

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

IMbackK committed 10 days ago

Verified 5f91b1d5

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

IMbackK committed 10 days ago

Verified 9ef7523e

Fix agentic mcp image single model (#20339)

ServeurpersoCom committed 10 days ago

Verified 00de6153

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

cabelo committed 11 days ago

Verified e1a39999

vendor : update miniaudio to 0.11.25 (#20209)

cabelo committed 11 days ago

Verified 4f2f0a16

fix op rope, add rope_back (#20293)

arthw committed 11 days ago

Verified 0cec84f9

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

arthw committed 11 days ago

Verified b2e1427c

model : qwen3vl reranker text support (#20332)

ViniciosLugli committed 11 days ago

Verified 4d99d450

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

ddh0 committed 11 days ago

Verified 10e5b148

ggml : bump RPC version (#20330)

ggerganov committed 11 days ago

Verified 90b27318

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

reeselevine committed 11 days ago

Verified aa2d278a

Reduce level of content parser warning message to avoid log spam on non-debug verbosity (#20347)

pwilkin committed 11 days ago

Verified 6c770d16

examples : fix empty items in json_schema_to_grammar.py [no ci] (#19968)

RayXu14 committed 11 days ago

Verified 8d880ac0

docs: update CPU backend ops to mark POOL_1D as supported (#20304)

a3894281 committed 11 days ago

Verified 0f1e9d14

models : fix assert in mamba2 (cont) (#20335)

ggerganov committed 11 days ago

Verified 1274fbee

server : make 2 checkpoints near the end of the prompt (#20288)

ggerganov committed 11 days ago

Verified a7b3dee7

common : fix incorrect uses of stoul (#20313)

CISC committed 11 days ago

Verified ec947d2b

kleidiai : support for concurrent sme and neon kernel execution (#20070)

chaxu01 committed 11 days ago

Verified 0cd4f472

ggml-cpu: add RVV repack GEMM and GEMV for quantization types (#19121)

taimur-10x committed 11 days ago

Verified af237f30

metal: handle command buffer failures gracefully in synchronize (#20306)

JulianPscheid committed 11 days ago

Verified 1a5631be

llama-quant : fail early on missing imatrix, refactor type selection, code cleanup (#19770)

ddh0 committed 11 days ago

Verified 1dab5f5a

common: consolidate PEG string parsers (#20263)

aldehir committed 12 days ago

Verified c96f608d

Older