ggml-org/llama.cpp

Pull Requests Commits

mpi : fix after master merge

ggerganov committed 2 years ago

Verified 04923631

Merge branch 'mpi' into refactor-mpi

ggerganov committed 2 years ago

Verified 81c5ddd5

[mpi] continue-on-error: true

evanmiller committed 2 years ago

03cc12be

Add OpenMPI to GH action

evanmiller committed 2 years ago

4a9a4748

Merge branch 'master' into mpi

evanmiller committed 2 years ago

0f557c2a

ggerganov committed 2 years ago

Verified 9da9d26c

mpi : fix inference

ggerganov committed 2 years ago

Verified beadbf33

mpi : fix output tensor after MPI compute (still not working)

ggerganov committed 2 years ago

Verified ef37dd14

mpi : various fixes - communication now works but results are wrong

ggerganov committed 2 years ago

Verified c717c518

mpi : move all MPI logic into ggml-mpi

ggerganov committed 2 years ago

Verified 01abb3b3

mpi : add names for layer inputs + prep ggml_mpi_graph_compute()

ggerganov committed 2 years ago

Verified e339d355

mpi : trying to move more MPI stuff into ggml-mpi (WIP) (#2099)

ggerganov committed 2 years ago

Verified 3232db62

llama : remove "first token must be BOS" restriction (#2153)

oobabooga committed 2 years ago

Verified 1d163099

main : escape prompt prefix/suffix (#2151)

pnb committed 2 years ago

Verified db4047ad

readme : update Termux instructions (#2147)

JackJollimore committed 2 years ago

Verified 18780e0a

ggml : fix buidling with Intel MKL but ask for "cblas.h" issue (#2104) (#2115)

clyang committed 2 years ago

Verified 3bbc1a11

readme : add more docs indexes (#2127)

rankaiyx committed 2 years ago

Verified 2492a53f

Fixed OpenLLaMA 3b CUDA mul_mat_vec_q (#2144)

JohannesGaessler committed 2 years ago

Verified 64639555

CUDA: add __restrict__ to mul mat vec kernels (#2140)

JohannesGaessler committed 2 years ago

Verified 061f5f8d

docker : add support for CUDA in docker (#1461)

canardleteer committed 2 years ago

Verified 84525e79

ci : switch threads to 1 (#2138)

ggerganov committed 2 years ago

Verified a7e20edf

ggml : change ggml_graph_compute() API to not require context (#1999)

mqy committed 2 years ago

Verified 1d656d63

ggml : remove sched_yield() call in ggml_graph_compute_thread() (#2134)

ggerganov committed 2 years ago

72421402

convert.py: add mapping for safetensors bf16 (#1598)

akx committed 2 years ago

Verified 3e08ae99

Add info to README

evanmiller committed 2 years ago

ef61acfb

Fix opencl by wrap #if-else-endif with \n (#2086)

howard0su committed 2 years ago

Verified 481f793a

Add GH workflow, fix test

evanmiller committed 2 years ago

55207ba2

Update CMakeLists.txt

evanmiller committed 2 years ago

1f0a2cfe

evanmiller committed 2 years ago

06a23934

Merge branch 'master' into mpi

evanmiller committed 2 years ago

32deabfd

Older