ggml-org/llama.cpp

Pull Requests Commits

fix partial writes

0cc4m committed 13 days ago

938872e9

0cc4m committed 13 days ago

ff6ad609

deduplicate repacking code

0cc4m committed 13 days ago

13a55c8e

add q4_1, q8_0, iq4_nl repacking

0cc4m committed 13 days ago

57fb74fb

replace malloc/free with thread_local memory

0cc4m committed 13 days ago

6906f781

add missing repacking functions

0cc4m committed 13 days ago

b64f294c

add mxfp4 repacking

0cc4m committed 13 days ago

b4e2621d

fix double semicolon

0cc4m committed 13 days ago

b1243aa9

add coopmat2 support

0cc4m committed 13 days ago

5c1e95c9

vulkan: repack q4_0 into aligned arrays

0cc4m committed 13 days ago

c285bb98

Support for Codex CLI by skipping unsupported Responses tools (#23041)

SidShaytay committed 13 days ago

Verified 91e84fed

readme : update bindings (#23063)

KitaitiMakoto committed 13 days ago

Verified 7155a497

ggml-hexagon: cpy: add contiguous fast-path in reshape copy (#23076)

pdhinaka committed 13 days ago

Verified 5c0e9468

HIP: RDNA3 mma FA, faster AMD transpose, tune AMD (#22880)

JohannesGaessler committed 14 days ago

Verified 3e037f31

CI : support IOT device (IQ9) (#22987)

zhiyuan8 committed 14 days ago

Verified d81e63dc

ggml-webgpu: Enable NVIDIA self-hosted CI (#22976)

reeselevine committed 14 days ago

Verified 834a2436

ggml-webgpu: makes the flash attn vec path subgroup-aware (#23040)

ArberSephirotheca committed 14 days ago

Verified 5ec717d1

fix: Propagate version tag to WebUI asset download in self-hosted CI (#23051)

allozaur committed 14 days ago

Verified 0c3e4fcc

contributing: new contributors should not submit trivial fixes (#23045)

am17an committed 14 days ago

Verified 97b658ce

webui: Move static build output from repo code to HF Bucket (#22937)

allozaur committed 14 days ago

Verified 253ba110

logs : reduce (#23021)

ggerganov committed 14 days ago

Verified 67b2b7f2

ggml-cpu: Add IME2 Instruction Support for the SpacemiT Backend (#22863)

alex-spacemit committed 14 days ago

Verified 81b0d882

docker : revert stable version of intel compute-runtime (#22968)

arthw committed 14 days ago

Verified 0f45f1a3

unicode,test: add Qwen3.5 non-backtracking tokenizer handler and regr… (#22110)

Kabir08 committed 14 days ago

Verified 42532aff

vulkan: fix matmul integer pipeline selection (#23005)

0cc4m committed 14 days ago

Verified dbe7901c

fix: Autoscroll detection (#23026)

allozaur committed 14 days ago

Verified 320a6a44

SYCL: fix multi-GPU system RAM exhaustion by using Level Zero allocations (#21597)

PMZFX committed 14 days ago

Verified 9ed6e19b

ggml-webgpu: only use subgroup-matrix path when head dims are divisible by sg_mat_k / sg_mat_n (#23020)

ArberSephirotheca committed 15 days ago

Verified 4c1c3ac0

Fix for issue #22974. Cast intermediate results to float before adding and casting the result to the destination type. Avoids half+half operator ambiguity. (#22994)

scutler-nv committed 15 days ago

Verified 7f3f843c

opencl: add q5_0 and q5_1 MoE for Adreno (#22985)

shaofeiqi committed 15 days ago

Verified ec562eb6

Older