ggerganov/llama.cpp

Pull Requests Commits

Merge branch 'master' into compilade/convert-hf-refactor

mofosyne committed 1 year ago

Verified c32d39ce

minor : fix trailing whitespace

ggerganov committed 1 year ago

Verified bcdee0da

Adding support for the --numa argument for llama-bench. (#7080)

kunnis committed 1 year ago

Verified 628b2991

Disable benchmark on forked repo (#7034)

CISC committed 1 year ago

Verified 8f8acc86

readme : add note that LLaMA 3 is not supported with convert.py (#7065)

lyledean1 committed 1 year ago

Verified ca363260

command-r : add BPE pre-tokenization (#7063)

dranger003 committed 1 year ago

Verified 889bdd76

py : logging and flake8 suppression refactoring (#7081)

mofosyne committed 1 year ago

Verified 6fbd4322

convert-hf : fix Refact conversion

compilade committed 1 year ago

215a0d38

gguf-split: add --no-tensor-first-split (#7072)

ngxson committed 1 year ago

Verified 84250014

Tidy Android Instructions README.md (#7016)

Jeximo committed 1 year ago

Verified cf768b7e

Fix Linux /sys cpu path to guess number of cores (#7064)

viric committed 1 year ago

Verified fcd84a0f

convert-hf : align the message logged for converted tensors

compilade committed 1 year ago

f2099c50

If first token generated from the server is the stop word the server will crash (#7038)

maor-ps committed 1 year ago

Verified 03fb8a00

tests : add test-tokenizer-0.sh + fix some tokenizers (#7036)

ggerganov committed 1 year ago

Verified 92139b90

convert-hf : more consistent formatting of cmdline args

compilade committed 1 year ago

98f2d0e0

Merge branch 'master' into compilade/convert-hf-refactor

compilade committed 1 year ago

3e5e0dce

convert.py : add python logging instead of print() (#6511)

mofosyne committed 1 year ago

Verified a2ac89d6

llama : rename ctx to user_data in progress_callback (#7045)

danbev committed 1 year ago

Verified 433def28

Merge branch 'master' into compilade/convert-hf-refactor

compilade committed 1 year ago

6a54973d

Remove .attention from skipped tensors to match more accurately (#7051)

bartowski1182 committed 1 year ago

Verified 60325fa5

convert-hf : use a plain class for Model, and forbid direct instantiation

compilade committed 1 year ago

13f4cf70

convert-hf : use an ABC for Model again

compilade committed 1 year ago

ce067af1

chore: fix typo in llama.cpp (#7032)

alwqx committed 1 year ago

Verified 6ecf3189

convert-hf : sort model part names

compilade committed 1 year ago

644c2696

convert-hf : convert norms to f32 by default

compilade committed 1 year ago

639b374b

Update LOG_IMPL and LOG_TEE_IMPL (#7029)

a-downing committed 1 year ago

Verified b0d943de

convert-hf : display tensor shape

compilade committed 1 year ago

21068b6b

main : fix off by one error for context shift (#6921)

l3utterfly committed 1 year ago

Verified 8d608a81

convert : use a string for the SentencePiece tokenizer path

compilade committed 1 year ago

dcd8dfa1

convert-hf : allow unusual model part names

compilade committed 1 year ago

3870164f

Older