ggml-org/llama.cpp

Pull Requests Commits

fix incorrect uses of stoul

CISC committed 7 hours ago

Verified 186b9075

server: Parse port numbers from MCP server URLs in CORS proxy (#20208)

eapache committed 10 hours ago

Verified 23fbfcb1

metal : extend mul_mv_ext to BF16, Q2_K, Q3_K (#20250)

arkavo-com committed 12 hours ago

Verified e22cd0aa

server : fix checkpoints n_tokens calculation (#20287)

ggerganov committed 12 hours ago

Verified 96cfc499

metal : add upscale (#20284)

ggerganov committed 12 hours ago

Verified ed0007aa

server : warn swa-full is not supported for non-SWA models (#20291)

ggerganov committed 12 hours ago

Verified 344ee2a3

server : fix off-by-1 in server_tokens::size_up_to_pos() (#20279)

ggerganov committed 12 hours ago

Verified d6e15564

common: map developer role to system (#20215)

pwilkin committed 13 hours ago

Verified f76565db

models : fix assert in mamba2 graph (#20270)

ggerganov committed 15 hours ago

Verified 43e1cbd6

server : add kill switch when server is stuck (#20277)

ggerganov committed 18 hours ago

Verified 107d5999

ggml-cuda: disable gdn for musa (#20278)

am17an committed 18 hours ago

Verified e8bbc736

llama-quant : left-align tensor names in output (#20117)

ddh0 committed 19 hours ago

Verified b5181951

contributing: limit open PRs for new contributors to 1 (#20036)

am17an committed 20 hours ago

Verified e2763a67

ggml-vulkan: add SGN operator, auto-generate Vulkan.csv and ops.md (#20219)

bertaye committed 20 hours ago

Verified 0beb8db3

vulkan: skip zero size tensors in backend copies (#20233)

0cc4m committed 20 hours ago

Verified b2f460bd

cuda : display total and free VRAM capacity during device initialization (#20185)

tehsiuhuang committed 22 hours ago

Verified 5f4cdac3

llama-bench: introduce `-hf` and `-hff` flags & use `--mmap 1` by default (#20211)

taronaeo committed 1 day ago

Verified ae87863d

PEG parser for LFM2 (#20251)

pwilkin committed 1 day ago

Verified 97c64fbd

server : do not create checkpoints right after mtmd chunks (#20232)

ggerganov committed 1 day ago

Verified d417bc43

graph : remove redundant scale_w parameter (#20235)

CISC committed 1 day ago

Verified 35bee031

common : gracefully handle incomplete output (#20191)

aldehir committed 1 day ago

Verified 451ef084

Fix compile bug (#20203)

pwilkin committed 1 day ago

Verified 9b24886f

Fix structured outputs (#20223)

pwilkin committed 1 day ago

Verified 62b8143a

ggml-vulkan: Add ELU op support (#20183)

GiantPrince committed 1 day ago

Verified d088d5b7

vulkan: Fix data races in coopmat1 mul_mat(_id) (#20084)

jeffbolznv committed 1 day ago

Verified cd18a50e

llama: end-to-end tests (#19802)

JohannesGaessler committed 1 day ago

Verified a976ff08

readme : update infra list (#20212)

Defilan committed 1 day ago

Verified a9504797

Revert to OAI-compatible args (#20213)

pwilkin committed 1 day ago

Verified b283f6d5

server : correct index on finish in OAI completion streams (#20226)

decahedron1 committed 1 day ago

Verified ff52ee96

[SYCL] supprt Flash Attention for fp32/fp16/Q4/Q5/Q8 (#20190)

arthw committed 1 day ago

Verified 213c4a0b

Older