ggml-org/llama.cpp

Pull Requests Commits

llama : remove redundant GQA check

ggerganov committed 1 year ago

Verified 7cfde781

examples : improve base-translate.sh script (#4783)

ggerganov committed 1 year ago

Verified 96e80dab

cmake : check for openblas64 (#4134)

a-n-n-a-l-e-e committed 2 years ago

Verified eec22a1c

flake.nix : fix typo (#4700)

eltociear committed 2 years ago

Verified be36bb94

metal : switch back to default.metallib (ggml/681)

ggerganov committed 2 years ago

91d38876

ggml : fix q2_k bpw in comments (ggml/680)

ggerganov committed 2 years ago

d061bf94

ggml : add error handling to graph_compute (whisper/1714)

finnvoor committed 2 years ago

1bf681f9

ggml : do not sched_yield when calling BLAS (#4761)

ggerganov committed 2 years ago

Verified c1d7cb28

examples : add few-shot translation example (#4783)

ggerganov committed 2 years ago

Verified 3681f224

finetune : remove unused includes (#4756)

danbev committed 2 years ago

Verified b3a7c20b

server : send token probs for "stream == false" (#4714)

ggerganov committed 2 years ago

Verified 012cf349

Print backend name on test-backend-ops failure (#4751)

JohannesGaessler committed 2 years ago

Verified a9192801

llama.swiftui : support loading custom model from file picker (#4767)

singularity-s0 committed 2 years ago

Verified 3c0b5855

server : fix options in README.md (#4765)

m18coppola committed 2 years ago

Verified e5804313

ggml : include stdlib.h before intrin.h (#4736)

ggerganov committed 2 years ago

Verified dc891b7f

llama.swiftui : fix build of ggml.metallib (#4754)

singularity-s0 committed 2 years ago

Verified 46cea79e

train : fix typo in overlapping-samples help msg (#4758)

danbev committed 2 years ago

Verified cb1e2818

swift : update Package.swift to use ggml as dependency (#4691)

1-ashraful-islam committed 2 years ago

Verified ece9a45e

cuda : simplify expression

ggerganov committed 2 years ago

7bed7eba

cuda : mark I16 and I32 ops as unsupported

ggerganov committed 2 years ago

d55356d3

ggerganov committed 2 years ago

75e3fd85

metal : add kernel_get_rows_i32

ggerganov committed 2 years ago

28931371

scripts : fix sync order + metal sed

ggerganov committed 2 years ago

ab62fc3e

ggml : extend ggml_get_rows, ggml_repeat, ggml_concat (ggml/639)

gwenzek committed 2 years ago

5f66ebca

server : throw an error when `slot unavailable` (#4741)

jparkerweb committed 2 years ago

Verified f2eb19bd

metal : optimize ggml_mul_mat_id (faster Mixtral PP) (#4725)

ggerganov committed 2 years ago

Verified f3f62f0d

server : add token counts to html footer (#4738)

phiharri committed 2 years ago

Verified 0ef3ca2a

llama : llama_model_desc print number of experts

ggerganov committed 2 years ago

Verified 540938f8

llama : replace all API facing `int`'s with `int32_t` (#4577)

MarcusDunn committed 2 years ago

Verified 0040d42e

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

postmasters committed 2 years ago

Verified 83e633c2

Older