LLM inference in C/C++
C++ ggml
Commits on master
common: add LOG_JSON macro to log structured data (#28586)
8e33095 Xuan-Son Nguyen committed 2 months ago
common : implement common_schema internal representation for JSON schemas (#28736)
acecd56 Aldehir Rojas committed 2 months ago
jinja : support dot property integer literals (#28817)
ae9afff Sigbjørn Skjæret committed 2 months ago
cmake: leave the timestamp out of precompiled headers on clang (#28816)
737e098 Pascal committed 2 months ago
server : allow model downloads at model limit fix issue #26809 (#28530)
56381e4 MiaoMing Chen committed 2 months ago
ggml-cuda: hip add specific config table for AMD GCN (#27841)
c8edceb thelittlefireman committed 2 months ago
vendor : update cpp-httplib to 0.56.0 (#28787)
718f7b4 Alessandro de Oliveira Faria (A.K.A.CABELO) committed 2 months ago
syscl : Handle (fail gracefully) unsupported tq1_0 quants (#28681)
2a3005c Michael Taylor committed 2 months ago
rpc : fix linking when compiling with BUILD_SHARED_LIBS=OFF (#28492)
f3a33df Ed Addario committed 2 months ago
server: frame the router child state command as a whole line (#28747)
c069aa7 Pascal committed 2 months ago
opencl: fix several bugs where the backend aborts (#27630)
8a56aed Hongqiang Wang committed 2 months ago
opencl: add bin kernel `kernel_gemm_noshuffle_q4_k_f32_32b_trans_ila_a8_bin` (#28677)
07fc977 shaofeiqi committed 2 months ago
hexagon: support for multi-device model split (aka row-split) (#28589)
eafe15a Max Krasnyansky committed 2 months ago
ggml-webgpu: Update to a recent version of Dawn (#28683)
d3146f2 Mendy Berger committed 2 months ago
cmake : skip PCH for llama-server PCH when using MSVC (#28763)
8ea2902 Daniel Bevenius committed 2 months ago
ci : run test-backend-ops as a dedicated ci/run.sh test (#28740)
b78a39a Georgi Gerganov committed 2 months ago
tests: extend test-quantize-fns to test nrc=2 (i8mm) kernels (#16234)
982937a Rohanjames1997 committed 2 months ago
tests: tolerate a shared pool abort in test_completion_unified (#28759)
8172e65 Pascal committed 2 months ago
ggml: skip 0-sized ids tensor when offloading selected experts (#28739)
43f3dda Aman Gupta committed 2 months ago
metal : skip the empty half of the mul_mm_id token tile (#28301)
5bda51b Foad Abo Dahood committed 2 months ago
cmake : add PCH and unity build to improve build times (#28091)
3bcfeb7 Daniel Bevenius committed 2 months ago
common : fix typo in speculative.cpp comment [no ci] (#28750)
1dfe94e Daniel Bevenius committed 2 months ago
metal : single-source fusion table + fusion debug rework (#28164)
a2878d3 Georgi Gerganov committed 2 months ago
metal : fix idle threads in the remaining iq mul_mv kernels for ne00 < 1024 (#28692)
aac8102 Foad Abo Dahood committed 2 months ago
model : fix MTP context kv cache allocation for deepseek2, glm4moe, … (#28630)
5cdd3d1 Logan Chu committed 2 months ago