ggml-org/llama.cpp

Pull Requests Commits
main : switch input_noecho to input_echo to remove negation (#979)
deadprogram committed 3 years ago
Verified e2cd5069
ggml: add names to tensors (#1268)
slaren committed 3 years ago
Verified 2d099e51
Add git-based build information for better issue tracking (#1232)
DannyDaemonic committed 3 years ago
Verified f4cef87e
cuBLAS: refactor and optimize f16 mat mul performance (#1259)
slaren committed 3 years ago
Verified 58b367c2
llama : update stubs for systems without mmap and mlock (#1266)
xloem committed 3 years ago
Verified ea3a0ad6
ggml : fix ggml_used_mem() (#1264)
KerfuffleV2 committed 3 years ago
Verified 2bdc0964
llama : fix session load / save (#1263)
ggerganov committed 3 years ago
Verified 70269cae
cuBLAS: fall back to pageable memory if pinned alloc fails (#1233)
slaren committed 3 years ago
Verified b925f1f1
llama : let context be const when accessing const data (#1261)
grencez committed 3 years ago
Verified 90b19bd6
ggml : fix UB (int << 31)
ggerganov committed 3 years ago
Verified 7ff0dcd3
build: add armv{6,7,8} support to cmake (#1251)
prusnak committed 3 years ago
Verified 6f796992
common : better default number of threads (#934)
jon-chuang committed 3 years ago
Verified a5d30b1f
ggml : add CLBlast q5_0, q5_1, q8_0 dequant kernels (#1225)
0cc4m committed 3 years ago
Verified 76a88492
ggml : add Q5 WASM SIMD + GGML_FTYPE
ggerganov committed 3 years ago
Verified 6bc4400e
Various fixes to mat_mul benchmark (#1253)
sw committed 3 years ago
Verified f0d70f14
ggml : fix labels for GGML_OP_ALIBI
ggerganov committed 3 years ago
Verified 3e5aa8a1
ggml : fix 32-bit ARM NEON
ggerganov committed 3 years ago
Verified c3ca7a5f
ggml : use vzip instead of vuzp for consistency
ggerganov committed 3 years ago
Verified e8c05161
ggml : fix visibility and unused warnings
ggerganov committed 3 years ago
Verified 0b5a9350
ggml : fix #if for f32_f32 mul_mat (CLBlast) (#1229)
ggerganov committed 3 years ago
Verified ec728e44
ggml : adjust mul_mat_f16 work memory (#1226)
ggerganov committed 3 years ago
Verified 214b6a35
build : fix reference to old llama_util.h
ggerganov committed 3 years ago
Verified 305eb5af
examples : fix save-load-state + rename llama-util.h
ggerganov committed 3 years ago
Verified 84ca9c2e
common : change default parameters to pre-#1126 (#1223)
ggerganov committed 3 years ago
Verified 334637e4
llama : new sampling algorithms (#1126)
ivanstepanovftw committed 3 years ago
Verified dd7eff57
cuBLAS: use host pinned memory and dequantize while copying (#1207)
slaren committed 3 years ago
Verified 7fc50c05
cuBLAS: non-contiguous tensor support (#1215)
SlyEcho committed 3 years ago
Verified b1ee8f59
Remove Q4_3 which is no better than Q5 (#1218)
sw committed 3 years ago
Verified 36d19a60
readme : update hot topics
ggerganov committed 3 years ago
Verified 7f15c5c4
ggml : sync ggml (ggml_alibi)
ggerganov committed 3 years ago
Verified 55390bca
Terms Of Use Privacy Policy Refund Policy Impressum