ggml-org/llama.cpp

Pull Requests Commits

ggml : fix rope args order + assert (#2054)

ggerganov committed 2 years ago

Verified 513f8619

gitignore : fix final newline

ggerganov committed 2 years ago

Verified 3973b25a

llama : remove cfg smooth factor as it is only a reparameterization of the guidance scale (#2280)

Vermeille committed 2 years ago

Verified ab0e26bd

gitignore : changes for Poetry users + chat examples (#2284)

yukiteruamano committed 2 years ago

Verified 73643f5f

make : fix indentation

ggerganov committed 2 years ago

Verified a814d04f

ci : fix MNT realpath usage (#2250)

ggerganov committed 2 years ago

Verified 4c013bb7

make : support customized LLAMA_CUDA_NVCC and LLAMA_CUDA_CCBIN (#2275)

skyan committed 2 years ago

Verified 42c7c2e2

flake : remove intel mkl from flake.nix due to missing files (#2277)

Freed-Wu committed 2 years ago

Verified 78a3d134

llama : make tensor_split ptr instead of array (#2272)

ggerganov committed 2 years ago

Verified ae178ab4

make : add new target for test binaries (#2244)

jpodivin committed 2 years ago

Verified 54e3bc76

MIKU MAYHEM: Upgrading the Default Model for Maximum Fun 🎉 (#2287)

at8u committed 2 years ago

Verified 019fe257

Faster Q2_K on Metal (#2297)

ikawrakow committed 2 years ago

Verified e68c96f7

make : fix embdinput library and server examples building on MSYS2 (#2235)

przemoc committed 2 years ago

Verified 9cf022a1

Faster Q5_K and Q6_K on Metal (#2294)

ikawrakow committed 2 years ago

Verified e782c9e7

Faster Q4_K on Metal (#2290)

ikawrakow committed 2 years ago

Verified 785829df

llama : fix regression from #2000 - could not load no-mmap models

ggerganov committed 2 years ago

fff0e0ea

metal: minor q4 optimization and reduce code size (#2248)

lshzh-ww committed 2 years ago

Verified 417a85a0

llama : extend API to get max devices at runtime (#2253)

SanftMonster committed 2 years ago

Verified 294f4245

flake : update flake.nix (#2270)

Freed-Wu committed 2 years ago

Verified 45a1b07e

cmake : install targets (#2256)

Freed-Wu committed 2 years ago

Verified b1f42909

ci : integrate with ggml-org/ci (#2250)

ggerganov committed 2 years ago

Verified d01bccde

llama : shorten quantization descriptions

ggerganov committed 2 years ago

Verified 6cbf9dfb

Support dup & cont ops on CUDA (#2242)

li-plus committed 2 years ago

Verified 7568d1a2

llama : fix t_start_sample_us initialization warning (#2238)

grencez committed 2 years ago

Verified b7647436

ggml : fixed runtime bugs and compile errors related to GGML_PERF and GGML_DEBUG (#2219)

mqy committed 2 years ago

Verified 672dda10

py : turn verify-checksum-models.py into executable (#2245)

jpodivin committed 2 years ago

Verified 27ab66e4

llama : add custom RoPE (#2054)

jxy committed 2 years ago

Verified 6e7cca40

flake : add runHook preInstall/postInstall to installPhase so hooks function (#2224)

ddellacosta committed 2 years ago

Verified a6803cab

make : use pkg-config for OpenBLAS (#2222)

Freed-Wu committed 2 years ago

Verified 7dabc66f

cuda : allocate all temporary ggml_tensor_extra_gpu from a fixed-size buffer (#2220)

bullno1 committed 2 years ago

Verified 7cdd30bf

Newer Older