ngxson/llama.cpp

Pull Requests Commits

llama: fix ASAN error with M-RoPE

ngxson committed 71 days ago

07d06fdd

llama: store mrope data in KV cell (#16825)

ngxson committed 71 days ago

Verified e3af5563

vulkan: Update topk_moe fusion to handle gpt's late softmax (#16656)

jeffbolznv committed 71 days ago

Verified 10fcc412

Vulkan MMQ Integer Dot Refactor and K-Quant support (#16536)

0cc4m committed 71 days ago

Verified bcf5bda6

Hexagon Op queue & dispatch optimizations (#16820)

max-krasnyansky committed 71 days ago

Verified 3eb2be1c

CUDA: use fastdiv in set-rows (#16834)

am17an committed 71 days ago

Verified e41bcce8

vendor : sync minja (#16500)

CISC committed 71 days ago

Verified 144a4ce8

vulkan: Call ggml_vk_buffer_write_2d from ggml_vk_buffer_copy (#16793)

jeffbolznv committed 72 days ago

Verified f549b000

CUDA: Fix bug in topk-moe for gpt-oss (#16821)

am17an committed 72 days ago

Verified 9a3ea685

sycl: add RMS_NORM_BACK operation support (#16808)

YaelLogic committed 72 days ago

Verified 338074c3

cuda: add SET operation support (#16804)

YaelGitAccount committed 72 days ago

Verified 851553ea

memory : remove KV cache size padding (#16812)

ggerganov committed 72 days ago

Verified 85a7d867

llama-bench : clarify benchmarked parts of the computation (#16823)

ggerganov committed 72 days ago

Verified a8ca18b4

initialise buffer.device in ggml_hexagon_session (#16816)

l3utterfly committed 72 days ago

Verified 8284efc3

embedding: add raw option for --embd-output-format (#16541)

SamMalayek committed 73 days ago

Verified 1c1409e1

llama: consistent ctx <-> buf order for KV cache (#16746)

JohannesGaessler committed 73 days ago

Verified 7a0e900e

grammar : support array references in json schema (#16792)

aldehir committed 73 days ago

Verified 280d97be

CANN: Improve device ID handling and aclnnArange checks (#16752)

noemotiovon committed 73 days ago

Verified 3479efd1

CUDA: add unused vars to mmvf and mmvq (#16807)

am17an committed 73 days ago

Verified 463bbf20

sycl: add SSM_CONV operation support (#16800)

tamarPal committed 73 days ago

Verified ad8d36be

chat: Add LFM2 tool handling (#16763)

ykhrustalev committed 73 days ago

Verified c053e18a

mtmd : fix idefics3 preprocessing (#16806)

ngxson committed 73 days ago

Verified e1ab0848

llama : disable pipeline parallelism if compute buffer allocation fails (#16748)

slaren committed 73 days ago

Verified 5a4ff43e

ggml : fix interpolate with align-corners and ne=1 (#16700)

Acly committed 73 days ago

Verified 10640e31

HIP: fix AMDGPU_TARGETS, update documentation (#16803)

JohannesGaessler committed 73 days ago

Verified 80d28f10

model : add LightOnOCR-1B model (#16764)

ngxson committed 73 days ago

Verified c55d53ac

llama: fix leaked buffers for mmap + split files (#16765)

JohannesGaessler committed 74 days ago

Verified 945501f5

test-backend-ops: print failed tests at the end (#16785)

am17an committed 74 days ago

Verified 75cbdd3f

sycl: add ROLL operation support (#16665)

tamarPal committed 74 days ago

Verified 2b9bd9bf

sycl: add REPEAT_BACK operation support (#16734)

shani-f committed 74 days ago

Verified 59fc1ec8

Older