ngxson/llama.cpp

Pull Requests Commits

ci : Use ggml-org/ccache-action on RISC-V as well (#21632)

luhenry committed 178 days ago

Verified 8612ed18

[SYCL] Fix Q8_0 reorder: garbage on 2nd prompt + crash on full VRAM (#21638)

PMZFX committed 178 days ago

Verified b1be68e8

server: use random media marker (#21962)

ngxson committed 178 days ago

Verified 408225bb

vulkan: optimize im2col (#21713)

0cc4m committed 179 days ago

Verified b3d75875

cuda: Q1_0 initial backend (#21629)

khosravipasha committed 179 days ago

Verified 7e72b38b

ggml-webgpu: Fix dequantization helpers to not pass in pointers (#21872)

reeselevine committed 179 days ago

Verified 20d3bc2c

CUDA: require explicit opt-in for P2P access (#21910)

JohannesGaessler committed 179 days ago

Verified a6206958

CUDA: manage NCCL communicators in context (#21891)

JohannesGaessler committed 179 days ago

Verified 014dca49

rpc : add native RDMA transport for RPC backend (RoCEv2) (#20590)

dvv101111 committed 179 days ago

Verified adb541a6

docs: more extensive RoPE documentation [no ci] (#21953)

ngxson committed 179 days ago

Verified 80d87708

ci: disable test-backend-ops on Vulkan llvmpipe run and resture default timeout (#21901)

0cc4m committed 179 days ago

Verified 8dc530b8

autoparser: support case of JSON_NATIVE with per-call markers (test case: Reka-Edge) (#21892)

pwilkin committed 179 days ago

Verified e1a9a6dc

read n_ctx back after making llama_context (#21939)

smashedpumpkin committed 179 days ago

Verified e39eba26

hexagon: optimization for HMX mat_mul (#21554)

njsyw1997 committed 179 days ago

Verified 5d14e5d1

ggml : remove ggml-ext.h (#21869)

ngxson committed 180 days ago

Verified fae3a280

metal : fix FA support logic (#21898)

ggerganov committed 180 days ago

Verified c0de6eda

mtmd: add mtmd_image_tokens_get_decoder_pos() API (#21851)

ngxson committed 180 days ago

Verified 707c0b7a

vulkan: Programmatically add RoundingModeRTE to all shaders when the device supports it (#21572)

jeffbolznv committed 180 days ago

Verified 1f30ac0c

ci : re-enable mac workflows (#21894)

ggerganov committed 180 days ago

Verified f4b5bf2f

metal : add XIELU unary op (#20802)

seyoungjeong committed 180 days ago

Verified aa0f1897

vendor : update BoringSSL to 0.20260413.0 (#21881)

angt committed 180 days ago

Verified be76dd0b

ggml : fix ARM NEON nvfp4 dot product on non-dotprod targets (#21559)

richarddd committed 180 days ago

Verified 2e05f06f

cmake: fix CMP0194 warning on Windows with MSVC (#21630)

texasich committed 180 days ago

Verified acc37a42

ggml-webgpu: Update register tiling matmul to use f32 accumulation (#21644)

reeselevine committed 180 days ago

Verified 5a23695d

common: skip reasoning budget sampler when no budget is requested (#21870)

berkidem committed 180 days ago

Verified 56666fa6

vulkan: Support GGML_TYPE_NVFP4 (#21455)

jeffbolznv committed 180 days ago

Verified 6a6780a2

server: support OAI /v1/audio/transcriptions API (#21863)

ngxson committed 180 days ago

Verified e489a5ca

common/gemma4 : handle parsing edge cases (#21760)

aldehir committed 180 days ago

Verified e21cdc11

docs: listing qwen3-asr and qwen3-omni as supported (#21857)

ngxson committed 180 days ago

Verified e9749236

chat: dedicated DeepSeek v3.2 parser + "official" template (#21785)

pwilkin committed 180 days ago

Verified 1c0d9081

Newer Older