ggerganov/llama.cpp

Pull Requests Commits

llama: various bug fixes

ngxson committed 25 days ago

c867e1c3

hexagon: fix Windows crash when op_poll is enabled (#26029)

adgup committed 25 days ago

Verified c0bc8591

CUDA: fix external compilation of q1_0 MMQ (#25778)

JohannesGaessler committed 25 days ago

Verified 1425386f

args: refactor mlock/mmap/directio into load-mode (#20834)

taronaeo committed 25 days ago

Verified e6dd0e29

contrib: fix leftovers from the AI usage policy update (#26030)

ServeurpersoCom committed 25 days ago

Verified da296d6e

metal : add f16 type support to leaky relu (#25981)

iliailmer committed 26 days ago

Verified c588c4f4

conversion: fix non-MoE NomicBert GGUF conversion error (#25996)

aorko01 committed 26 days ago

Verified d941f6e1

contrib: allow all AI-generated code in general (#26012)

ngxson committed 26 days ago

Verified 4310aa4f

ui: Add a "Default" option for the reasoning selector (#25846)

ServeurpersoCom committed 26 days ago

Verified cf512566

CUDA: Improve NVFP4 W4A4 activation quantization (#25730)

ORippler committed 26 days ago

Verified 1a064ab0

hexagon: activation ops update (#25974)

tboinovski1 committed 26 days ago

Verified 0278d836

mtmd: use RAII for setting and resetting non-causal attention (#25723)

nikwen committed 26 days ago

Verified e0833bf6

feat(ui): add symbolic math support to JS sandbox via nerdamer (#25948)

rankaiyx committed 26 days ago

Verified 61328e6a

minor: fix reasoning preserve var for DS4 [no ci] (#25999)

pwilkin committed 26 days ago

Verified e8e6c7af

common: infer the speculative type from the draft repo sidecars (#25989)

ServeurpersoCom committed 26 days ago

Verified 6d5a910c

Fix DeepSeek4 crafted template (#25414)

pwilkin committed 26 days ago

Verified f534da26

ggml: enable PowerPC backend variants on AIX (#25983)

shalinib-ibm committed 26 days ago

Verified 3ce7da2c

ci : fix SYCL package shared library lookup (#25987)

KyleHagy committed 26 days ago

Verified b4d6c7d8

webgpu : add CONV_2D_DW (depthwise conv2d) kernel (#25847)

m1el committed 26 days ago

Verified 7347430f

cuda: GET_ROWS quants (#25962)

ServeurpersoCom committed 27 days ago

Verified c5a4a0bb

llama-arch: fix DeepSeek4 APE tensor op (#25945)

helanfxz committed 27 days ago

Verified 67b9b0e7

Add support for Laguna XS.2 & M.1 (#25165)

joerowell committed 27 days ago

Verified 1f66c3ce

convert: fix handle HunyuanVL XD-RoPE config (#25514)

wendadawen committed 27 days ago

Verified 66e4bf7e

mtmd : use align_corners for qwen3vl vision position embedding interpolation (#25781)

gerbenvv committed 27 days ago

Verified b4aa7dd4

hexagon: check tensor type when reusing descriptors (#25968)

w1049 committed 27 days ago

Verified 71102a73

cuda: add sqrt_softplus in topk-moe for dsv4 (#25896)

am17an committed 27 days ago

Verified 846e991e

kleidiai : warn once when a weight type has no KleidiAI kernel (#25701)

kkjjkamal123 committed 27 days ago

Verified fb0e6b62

common: resolve draft repo to its requested sidecar (#25955)

ServeurpersoCom committed 27 days ago

Verified 60f6a177

server: return 400 instead of 500 on validation error with X-Conversation-Id (#25760)

ServeurpersoCom committed 27 days ago

Verified fd41bf65

server : properly handle null llama_context (#25868)

fairydreaming committed 27 days ago

Verified 40b740ad

Older