ngxson/llama.cpp

Pull Requests Commits

remove attn_mask

ngxson committed 47 days ago

7161371d

re-order the set_input case

ngxson committed 47 days ago

00fb9ea6

correct audio tokens

ngxson committed 47 days ago

7117f29e

limit mtmd_chunk size (avoid blow up memory)

ngxson committed 48 days ago

1e8a82ec

ngxson committed 48 days ago

98b2af7a

mtmd: add chunks and fix preproc for qwen3a

ngxson committed 48 days ago

5031aef5

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

reeselevine committed 48 days ago

Verified 834a2436

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

ArberSephirotheca committed 48 days ago

Verified 5ec717d1

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

allozaur committed 48 days ago

Verified 0c3e4fcc

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

am17an committed 48 days ago

Verified 97b658ce

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

allozaur committed 48 days ago

Verified 253ba110

logs : reduce (#23021)

ggerganov committed 48 days ago

Verified 67b2b7f2

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

alex-spacemit committed 48 days ago

Verified 81b0d882

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

arthw committed 48 days ago

Verified 0f45f1a3

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

Kabir08 committed 48 days ago

Verified 42532aff

vulkan: fix matmul integer pipeline selection (#23005)

0cc4m committed 48 days ago

Verified dbe7901c

fix: Autoscroll detection (#23026)

allozaur committed 48 days ago

Verified 320a6a44

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

PMZFX committed 48 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 49 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 49 days ago

Verified 7f3f843c

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

shaofeiqi committed 49 days ago

Verified ec562eb6

server, webui: accept continue_final_message flag for vLLM API compat (#23012)

ServeurpersoCom committed 49 days ago

Verified 95d469a9

opencl: fix crash when warming up MoE on Adreno (#22876)

lhez committed 49 days ago

Verified 1e4579fb

flush the gpu profile timestamp before the queryset is overflowed (#22995)

yomaytk committed 49 days ago

Verified 527045bf

webui: Deduplicate model aliases in data + handle single/multiple aliases in UI (#22979)

allozaur committed 49 days ago

Verified 2dfeca31

webui: preserve system message on edit cancel (#22911)

ServeurpersoCom committed 49 days ago

Verified 46be24d1

docs : Update OPENVINO.md (#22959)

ravi9 committed 49 days ago

Verified 7e166460

hexagon: add unary tanh op (#22999)

max-krasnyansky committed 49 days ago

Verified ad96bb8c

download: do not exit() on error (#23008)

ngxson committed 49 days ago

Verified e75cd5ef

server, webui: support continue generation on reasoning models (#22727)

ServeurpersoCom committed 49 days ago

Verified 5d44db60

Older