ggerganov/llama.cpp

Pull Requests Commits

parallelize fattn compilation test

slaren committed 1 year ago

1ca802a3

fix nwarps > batch size

JohannesGaessler committed 1 year ago

f4003cfb

add q8_0 q4_0 tests

JohannesGaessler committed 1 year ago

f0877604

fix commented-out kernel variants

JohannesGaessler committed 1 year ago

3194a010

JohannesGaessler committed 1 year ago

462add6a

CUDA: quantized KV support for FA vec

JohannesGaessler committed 1 year ago

672244a8

make: add --device-debug to NVCC debug flags (#7542)

JohannesGaessler committed 1 year ago

Verified 10b1e458

Allow multiple copy function pointers for CUDA graph kernel param updates (#7565)

agray3 committed 1 year ago

Verified 197c0068

Fix q_xxs using mul_mat_q (#7459)

AidanBeltonS committed 1 year ago

Verified 95f84d5c

Add freq factors (#7495)

AidanBeltonS committed 1 year ago

Verified 5487593b

metal : add GGML_OP_REPEAT kernels (#7557)

ggerganov committed 1 year ago

Verified 1d8fca72

metal : disable FA kernel for HS=256 (#7556)

ggerganov committed 1 year ago

Verified 62bfef51

llama : add comments about experimental flags (#7544)

ggerganov committed 1 year ago

Verified eaf6e031

github: add self sorted issue ticket forms (#7543)

mofosyne committed 1 year ago

Verified d6ef0e77

flake.lock: Update (#7540)

ggerganov committed 1 year ago

Verified dff451cf

main: replace --no-special with --special (#7534)

mofosyne committed 1 year ago

Verified d298382a

Fix aya-23 conversion scripts (#7539)

Galunid committed 1 year ago

Verified 32a28217

llama : add Smaug 70B support (#7402)

bartowski1182 committed 1 year ago

Verified c429b33b

Readme: add akx/ggify to tools (#1484)

akx committed 1 year ago

Verified 9146d36f

SimpleChat Completion Mode flexibility and cleanup, Settings gMe, Optional sliding window (#7480)

hanishkvc committed 1 year ago

Verified b9adcbbf

train : change default FA argument (#7528)

ggerganov committed 1 year ago

Verified 9588f196

labeler: added Apple Metal detector (+Kompute) (#7529)

mofosyne committed 1 year ago

Verified 3cbd23ed

main : don't print special tokens with --grammar (#6923)

jart committed 1 year ago

Verified 00c63907

ggml: aarch64: SVE kernels for q8_0_q8_0, q4_0_q8_0 vector dot (#7433)

msy-kato committed 1 year ago

Verified faa0e697

android : module (#7502)

eltonkola committed 1 year ago

Verified 9791f402

fix missing slash in `fs_get_cache_directory()` (#7503)

ngxson committed 1 year ago

Verified 902184dd

Make tokenize CLI tool have nicer command line arguments. (#6188)

Noeda committed 1 year ago

Verified 57684331

gguf-py : fix and simplify quantized shape round-trip (#7483)

compilade committed 1 year ago

Verified b83bab15

flake.lock: Update (#7232)

ggerganov committed 1 year ago

Verified d041d2ce

docker.yml: disable light-intel and server-intel test (#7515)

mofosyne committed 1 year ago

Verified 27891f6d

Older