ggerganov/llama.cpp

Pull Requests Commits

cont : remove /api/tags

ggerganov committed 59 days ago

Verified 35df147d

server : remove /api endpoints

ggerganov committed 59 days ago

Verified c1891fd6

fix: GLM-DSA crash in llama-tokenize when using vocab_only (#22102)

ssam18 committed 59 days ago

Verified 81df3f7c

server : refactor "use checkpoint" logic (#22114)

ggerganov committed 59 days ago

Verified de71b5f8

[SYCL] Fix reorder MMVQ assert on unaligned vocab sizes (#22035)

PMZFX committed 59 days ago

Verified 788fcbc5

server: rename --clear-idle to --cache-idle-slots (#21741)

yychyo committed 59 days ago

Verified 9d49acb2

vendor : update cpp-httplib to 0.42.0 (#21781)

cabelo committed 59 days ago

Verified e365e658

CUDA: refactor mma data loading for AMD (#22051)

JohannesGaessler committed 59 days ago

Verified 4eac5b45

common/autoparser : allow space after tool call (#22073)

aldehir committed 60 days ago

Verified d5b780a6

HIP: Remove unesscary NCCL_CHECK (#21914)

IMbackK committed 60 days ago

Verified 471540ae

mtmd: add pos_0 to mtmd_image_tokens_get_decoder_pos (breaking change) (#22082)

ngxson committed 60 days ago

Verified 19124078

ggml : reduce CPU overhead in meta backend (#22041)

gaugarg-nv committed 60 days ago

Verified bcdcc104

ci : install spirv-headers for vulkan-cross (#22109)

CISC committed 60 days ago

Verified 037bfe38

convert : support sentence-transformer 5.4 config files (#22087)

Bing-su committed 60 days ago

Verified 8685e7b0

cmake: remove CMP0194 policy to restore MSVC builds (#21934)

texasich committed 60 days ago

Verified 09b4efa9

server : speculative checkpointing (#19493)

srogmann committed 60 days ago

Verified 455d8e4b

rpc : refactor the RPC transport (#21998)

rgerganov committed 60 days ago

Verified 91fef953

server: Expose `media_tag` on /props endpoint. (#22028)

cetarthoriphros committed 60 days ago

Verified 9e5647af

model : refactor bias tensor variable names (#22079)

CISC committed 60 days ago

Verified 4f02d473

android : libcommon -> libllama-common (#22076)

CISC committed 61 days ago

Verified 23b8cc49

ggml-backend-meta: add multi-segment read support in get_tensor (#22063)

ssam18 committed 61 days ago

Verified 59accc88

ci : free disk space for rocm release (#22012)

CISC committed 61 days ago

Verified 83d58e02

convert : fix (ignore for now) typings errors (#22002)

CISC committed 61 days ago

Verified 89a5474f

llama: fit ctx size for CPU only (#21568)

JohannesGaessler committed 61 days ago

Verified fd1c0ec3

ggml-webgpu: fix compiler warnings and refactor FlashAttention encoding (#21052)

reeselevine committed 61 days ago

Verified 45cac7ca

CUDA: use LRU based eviction for cuda graphs (#21611)

am17an committed 61 days ago

Verified b94050e8

ci : add android arm64 build and release (#21647)

ykhrustalev committed 62 days ago

Verified a279d0f0

mtmd: add missing struct tag (#22023)

65a committed 62 days ago

Verified 268d61e1

libs : rename libcommon -> libllama-common (#21936)

ggerganov committed 62 days ago

Verified 6990e2f1

model : Gemma4 model type detection (#22027)

EZForever committed 62 days ago

Verified fcc75087

Older