ggml-org/ggml

Pull Requests Commits

ggml : fix mul_mat src1 indexing when src1 is not contiguous

ggerganov committed 3 years ago

Verified c5a21cd2

ggml : apply mul_mat broadcast fix (sync llama.cpp)

ggerganov committed 3 years ago

Verified f6365c06

cuda : sync llama.cpp

ggerganov committed 3 years ago

Verified 56216523

ggml : broadcast mul_mat + conv batch support (#325)

monatis committed 3 years ago

Verified c2a5a358

ggml : pool naming changes + coding style fixes

ggerganov committed 3 years ago

Verified b2140642

ggml : add CUDA gelu op (#374)

li-plus committed 3 years ago

Verified 9ae9b781

ggml : basic implementation of 1d and 2d pools (#375)

iboB committed 3 years ago

Verified 25b2d4eb

ggml : revert CUDA broadcasting changes until fixed

ggerganov committed 3 years ago

Verified 8e2d6dfa

ggml : remove unused function (#328)

ggerganov committed 3 years ago

Verified 1b60ce00

ggml : fix compile warnings after #328

ggerganov committed 3 years ago

Verified b3f3dce1

ggml : add callback to abort ggml_graph_compute() (#328)

CCLDArjun committed 3 years ago

Verified aded8980

ggml : use a single kernel for CUDA mul op (#373)

li-plus committed 3 years ago

Verified f5165d03

tests : ifdef for #pragma GCC (#370)

iboB committed 3 years ago

Verified 0c55da24

readme : add link to ggllm.cpp repo (close #361)

ggerganov committed 3 years ago

Verified 69b440c0

ggml : broadcast ggml_add() for F32 (#359)

li-plus committed 3 years ago

Verified 649a922d

ggml : support norm op on CUDA (#364)

li-plus committed 3 years ago

Verified 17f675cb

starcoder : fix for ggml-model.bin being saved in wrong directory + use argparse (#363)

TheBloke committed 3 years ago

Verified b5829db7

ggml : fix ggml_set_xxx (#354)

goerch committed 3 years ago

Verified 61fe6df5

ggml : fix Alibi implementation (#351)

daulet committed 3 years ago

Verified 2c1536bd

ggml : sync llama.cpp (fix for #341)

ggerganov committed 3 years ago

Verified 3ea676a3

ggml : fix docs about element access (close #348)

ggerganov committed 3 years ago

Verified 9bc39eec

starcoder : add <|end_of_turn|> token handling in order to support openchat/opencoderplus (#343)

the-crypt-keeper committed 3 years ago

Verified d40d7ff5

pkg-config : fix typo in includedir (#367)

smspillaz committed 3 years ago

Verified ad2754ef

ggml : sync llama.cpp (changes to ggml_graph_compute() API) (#368)

ggerganov committed 3 years ago

Verified e4192481

ggml : minor indentation

ggerganov committed 3 years ago

Verified 543e8897

ggml : restore GGML_RESTRICT (#350)

iboB committed 3 years ago

Verified 4b41bae7

ggml : fix bug introduced in bfc6d42f8c2141383e4f21e4a030688c71560da0

ggerganov committed 3 years ago

Verified 43ffec5f

tests : sync from llama.cpp and disable some obsolete tests

ggerganov committed 3 years ago

Verified d8fbf15c

ggml : sync llama.cpp (generalize quantize_fns + CUDA improvements)

ggerganov committed 3 years ago

Verified bfc6d42f

cmake : fix public header path for submodules (#342)

abetlen committed 3 years ago

Verified d5c4ce0b

Older