ngxson/llama.cpp

Pull Requests Commits
ggml-cpu: aarm64: q4_K repack gemm and gemv implementations (dotprod only) (#17494)
Alcpz committed 47 days ago
Verified cd8370b4
devops: Add build-essential to Ubuntu 26.04 image (#17531)
ericcurtin committed 47 days ago
Verified d21a76ac
gguf-py : skip endian-conversion of MXFP4 data (#17523)
AlekseiNikiforovIBM committed 47 days ago
Verified 4fcd87cf
vulkan : move contiguous checks to device_supports_op (#17490)
Acly committed 47 days ago
Verified b78db3bd
vulkan: use a fixed 1KB buffer for the add_rms_fusion opt (#17514)
jeffbolznv committed 47 days ago
Verified 142df17c
server: enable jinja by default, update docs (#17524)
ngxson committed 47 days ago
Verified e509411c
opencl: add sqr, sqrt, mean and ssm_conv (#17476)
lhez committed 47 days ago
Verified 7cba58bb
Fix chunks being too small with small matrix sizes (#17526)
Alcpz committed 47 days ago
Verified 5449367b
clip: (minicpmv) fix resampler kq_scale (#17516)
hNSBQZ committed 47 days ago
Verified 1d594c29
vulkan: allow graph_optimize for prompt processing workloads (#17475)
jeffbolznv committed 48 days ago
Verified eec1e33a
vulkan: Implement top-k (#17418)
jeffbolznv committed 48 days ago
Verified 879d6737
ggml-cpu : add RISC-V Zvfh impl for ggml_vec_mad_f16 (#17448)
xctan committed 48 days ago
Verified 6ab4e50d
cmake : use EXCLUDE_FROM_ALL to avoid patch-boringssl.cmake (#17520)
angt committed 48 days ago
Verified 2336cc47
ggml : fix ARM feature verification (#17519)
angt committed 48 days ago
Verified e6923caa
HIP: Patch failed testcase in WMMA-MMQ kernels for RDNA 4 (#17502)
jiachengjason committed 48 days ago
Verified 3e18dba9
CANN: Add MROPE and IMROPE support (#17401)
hipudding committed 48 days ago
Verified eeb5605d
chore: upgrade cpp-httplib from v0.27.0 to v0.28.0 (#17513)
o7si committed 48 days ago
Verified f3a848a3
vulkan: Implement GGML_OP_CUMSUM (#17479)
jeffbolznv committed 48 days ago
Verified b3b03a7b
ggml : add ggml_top_k (#17365)
ggerganov committed 49 days ago
Verified 583cb834
convert : fix big-endian conversion (#17431)
AlekseiNikiforovIBM committed 49 days ago
Verified 05872ac8
codeowners : remove slaren (#17492)
slaren committed 49 days ago
Verified 55ab25ca
CANN: supports out_prod operator for F32 and F16 (#17406)
TianHao324 committed 49 days ago
Verified 064c90d8
webui: add rehype plugin to restore HTML in Markdown table cells (#17477)
ServeurpersoCom committed 49 days ago
Verified b1846f1c
vulkan: Use fewer rows for scalar FA when HS is not a multiple of 16 (#17455)
jeffbolznv committed 49 days ago
Verified d414db02
llama: introduce support for model-embedded sampling parameters (#17120)
taronaeo committed 49 days ago
Verified 877566d5
vulkan: more FA details in vk_perf_logger (#17443)
jeffbolznv committed 49 days ago
Verified 3d07caa9
llama : skip output reordering for single token batches (#17466)
danbev committed 49 days ago
Verified 134e6940
HIP: WMMA-MMQ kernels for RDNA 4 (#17156)
jiachengjason committed 49 days ago
Verified 0543f928
convert : allow quantizing lora again (#17453)
CISC committed 50 days ago
Verified b61de2b2
server: split server.cpp code into server/common/task/queue (#17362)
ngxson committed 50 days ago
Verified b8372eec
Terms Of Use Privacy Policy Refund Policy Impressum