ngxson/llama.cpp

Pull Requests Commits

ngxson committed 4 days ago

77864903

Merge pull request #72 from aldehir/jinja-vm-whitespace

ngxson committed 4 days ago

Verified 14a87064

adjust whitespace rules

aldehir committed 4 days ago

644d2815

fix testing.h path

aldehir committed 4 days ago

c0add06c

create jinja rendering tests

aldehir committed 4 days ago

90062624

more builtins added, more fixes

ngxson committed 5 days ago

7fbdf63d

improve error reporting

ngxson committed 5 days ago

2ca9d799

common/grammar : replace problematic backtracking regex `[\s\S]*` (#18342)

aldehir committed 5 days ago

Verified cef1d23c

fully functional selectattr / rejectattr

ngxson committed 5 days ago

264dcea2

graph : fix graph reuse logic when `n_pos_per_embd > 1` (#18566)

ggerganov committed 5 days ago

Verified c69c7ebc

add object.length

ngxson committed 5 days ago

4af1850b

make sure tojson output correct order

ngxson committed 5 days ago

13ddab2a

disable tests for COMMON_CHAT_FORMAT_GENERIC

ngxson committed 5 days ago

60a3a6a2

ngxson committed 5 days ago

8c01e0ec

ngxson committed 5 days ago

99aa61ca

correct lstrip behavior

ngxson committed 5 days ago

85b0efe5

improve error reporting

ngxson committed 5 days ago

25a884ec

ggml-cuda: fixes for concurrent streams (#18496)

am17an committed 5 days ago

Verified e57f5233

rename vm --> interpreter

ngxson committed 5 days ago

e392fef4

context : fix reserve token padding to n_seqs (#18536)

ggerganov committed 5 days ago

Verified a554a1ec

CUDA: only allocate FA tmp buffer if needed (#18564)

JohannesGaessler committed 5 days ago

Verified 0f2e42ca

ngxson committed 5 days ago

16e5d524

ngxson committed 5 days ago

cf521dc9

test-chat-template: use common

ngxson committed 5 days ago

8fb879b7

rm test-chat-jinja from cmake

ngxson committed 5 days ago

217afcd3

move test-chat-jinja into test-chat-template

ngxson committed 5 days ago

50aa8ed2

fix test-chat-template

ngxson committed 5 days ago

1de836b5

ngxson committed 5 days ago

9b24eadf

(Bugfix, ggml-cuda) Pool alloc count fix + small size computation type adjustment (#18559)

pl752 committed 5 days ago

Verified 9dba9f53

ggml-hexagon: optimize activation function (#18393)

joeldushouyu committed 6 days ago

Verified bcfc8c3c

Newer Older