ggml-org/llama.cpp

Pull Requests Commits

gguf-py : do not align the data start offset

compilade committed 23 hours ago

5f14aa8e

llama: fix RPC for -fit on (#18233)

JohannesGaessler committed 1 day ago

Verified 0e1ccf15

move copilot instructions to AGENTS.md (#18259)

ngxson committed 1 day ago

Verified 5e25ddeb

vulkan: fix im2col overflowing maxworkgroupcount (#18180)

jeffbolznv committed 2 days ago

Verified fd05c51c

vulkan/cuda: fix topk_moe with exp_probs_b (#18071)

jeffbolznv committed 2 days ago

Verified b365c3ff

vulkan: support GGML_UNARY_OP_XIELU (#18062)

jeffbolznv committed 2 days ago

Verified cb64222b

vulkan: in graph_optimize, try to group ADD operations (#18060)

jeffbolznv committed 2 days ago

Verified 6eb70818

Vulkan: some improvement on mul_mat_iq2_xs (#18031)

lovedheart committed 2 days ago

Verified 4117ae55

docs : fix links in parsing.md (#18245)

danbev committed 2 days ago

Verified 65e96a24

common : reorganize includes to prioritize vendored deps (#18222)

aldehir committed 2 days ago

Verified 9496bbb8

server: add auto-sleep after N seconds of idle (#18228)

ngxson committed 2 days ago

Verified ddcb75dd

tests: Avoid floating point precision false positives in SUM (#17471)

jeffbolznv committed 2 days ago

Verified 52ab19df

test-backend-ops: improve msvc build time (#18209)

jeffbolznv committed 2 days ago

Verified 5182dd64

Added comments explaining thread block size selection logic based on row count and column size, derived from historical commit context (#18212)

Aadeshveer committed 3 days ago

Verified 10b4f82d

server : [easy] fix per round speculative decode logging (#18211)

okuvshynov committed 3 days ago

Verified 408616ad

server: support load model on startup, support preset-only options (#18206)

ngxson committed 3 days ago

Verified 9e39a1e6

ci : remove non-windows zip artifacts (#18201)

CISC committed 3 days ago

Verified 74e05131

ci : only save ccache on master (#18207)

CISC committed 3 days ago

Verified f74747d8

ggml-hexagon: Implement true Q8_0 quantization on Hexagon NPU for more accurate mixed-precision matmul operations (#17977)

ngdxzy committed 3 days ago

Verified ce734a8a

arg: fix order to use short form before long form (#18196)

ServeurpersoCom committed 3 days ago

Verified 14931a82

llama : Changing off_t to size_t for Windows (#18204)

JTischbein committed 3 days ago

Verified f99ef53d

server: friendlier error msg when ctx < input (#18174)

am17an committed 4 days ago

Verified cc0a0434

presets: refactor, allow cascade presets from different sources, add global section (#18169)

ngxson committed 4 days ago

Verified 98c1c7a7

webui: Add editing attachments in user messages (#18147)

allozaur committed 4 days ago

Verified acb73d83

model-conversion : add verbose flag in run-org-model.py (#18194)

danbev committed 4 days ago

Verified 0a271d82

android: fix missing screenshots for Android.md (#18156)

naco-siren committed 4 days ago

Verified 52fc7fee

vulkan: Add perf logger mode with concurrency (#17944)

jeffbolznv committed 4 days ago

Verified cdbada8d

model : add ASR support for LFM2-Audio-1.5B (conformer) (#18106)

ngxson committed 4 days ago

Verified 8ea958d4

webui: display prompt processing stats (#18146)

ServeurpersoCom committed 4 days ago

Verified f9ec8858

ggml-cpu: extend support for RVV floating-point kernels (#17318)

taimur-10x committed 5 days ago

Verified f716588e

Older