ggml-org/llama.cpp

Pull Requests Commits

llama : refactor quantization to avoid <mutex> header

ggerganov committed 2 years ago

Verified b5af7ad8

llama : differentiate the KV dims in the attention (#4657)

postmasters committed 2 years ago

Verified 83e633c2

editorconfig : fix whitespace and indentation #4710

ggerganov committed 2 years ago

Verified 32866c5e

server : add --override-kv parameter (#4710)

minarchist committed 2 years ago

Verified 5d7002d4

py : re-enable mmap in convert hf (#4732)

namtranase committed 2 years ago

Verified 26f3071d

finetune: fix typo in README.md (#4733)

danbev committed 2 years ago

Verified 775ac871

metal : enable shader debugging (cmake option) (#4705)

ggerganov committed 2 years ago

Verified 58ba655a

flake.lock: update

SomeoneSerge committed 2 years ago

edd1ab7b

flake.nix: suggest the binary caches

SomeoneSerge committed 2 years ago

198ed7eb

workflows: nix-ci: add a qemu job for jetsons

SomeoneSerge committed 2 years ago

d8361747

workflows: nix-flakestry: drop tag filters

SomeoneSerge committed 2 years ago

06f2a5d1

workflows: weekly `nix flake update`

SomeoneSerge committed 2 years ago

c5239944

workflows: nix-ci: add a job for eval

SomeoneSerge committed 2 years ago

1e9ae54c

workflows: nix-ci: init; build flake outputs

SomeoneSerge committed 2 years ago

7adedecb

flake.nix: expose checks

SomeoneSerge committed 2 years ago

356ea17e

flake.nix: rocm not yet supported on aarch64, so hide the output

SomeoneSerge committed 2 years ago

a5c088d8

flake.nix: expose full scope in legacyPackages

SomeoneSerge committed 2 years ago

1e3900eb

ggml : add ggml_vdotq_s32 alias (#4715)

ggerganov committed 2 years ago

Verified e39106c0

clip : refactor + bug fixes (#4696)

ggerganov committed 2 years ago

Verified 9fbda719

CUDA: fixed tensor cores not being used on RDNA3 (#4697)

JohannesGaessler committed 2 years ago

Verified 39d8bc71

ggml : add ggml_cpu_has_avx_vnni() (#4589)

tikikun committed 2 years ago

Verified 24a447e2

CUDA: fix tensor core logic for Pascal and HIP (#4682)

JohannesGaessler committed 2 years ago

Verified a20f3c74

clip : use ggml_backend_buffer_is_host (#4205)

ggerganov committed 2 years ago

Verified 0235b9b5

clip : enable gpu backend (#4205)

FSSRepo committed 2 years ago

Verified ce18d727

cuda: fix vmm oom issue on NVIDIA AGX Orin (#4687)

hydai committed 2 years ago

Verified 91bb39ce

python : add check-requirements.sh and GitHub workflow (#4585)

crasm committed 2 years ago

Verified 04ac0607

flake.nix : rewrite (#4605)

philiptaron committed 2 years ago

Verified 68eccbdc

cmake : fix ld warning duplicate libraries libllama.a (#4671)

nguoithichkhampha committed 2 years ago

Verified 97bbca6e

llava-cli : refactor to use sampling library (#4669)

jart committed 2 years ago

Verified 4af48015

server : replace sleep with condition variables (#4673)

jart committed 2 years ago

Verified db49ff8e

Older