COMMITS
/ src/unicode.cpp April 7, 2026
S
unicode : add custom Qwen2 regex handler to fix segfault on long input (#21257)
Son H. Nguyen committed
April 4, 2026
A
llama: add custom newline split for Gemma 4 (#21406)
Aman Gupta committed
April 3, 2026
P
vocab: fix Gemma4 tokenizer (#21343)
Piotr Wilkin (ilintar) committed
March 5, 2026
M
chore : correct typos [no ci] (#20041)
Marcel Petrick committed
February 16, 2026
S
model: Add support for Tiny Aya Models (#19611)
Saurabh Dash committed
February 13, 2026
A
llama : remove deprecated codecvt (#19565)
Adrien Gallouët committed
February 6, 2026
L
unicode : MSVC regex fix (#19340)
Lasse Lauwerys committed
January 1, 2026
T
model: support youtu-vl model (#18479)
tt committed
December 5, 2025
P
November 14, 2025
B
model : add AfmoeForCausalLM support (#16477)
Bartowski committed
July 15, 2025
G
model : add Kimi-K2 support (#14654)
Gabriel Larson committed
June 19, 2025
F
build : suppress gcc15 compile warnings (#14261)
fanyang committed
February 15, 2025
G
repo : update links to new url (#11886)
Georgi Gerganov committed
February 7, 2025
C
vocab : ignore invalid UTF-8 input in the BPE tokenizer (#11729)
Christian Fillion committed
January 18, 2025
G
cmake : add sanitizer flags for llama.cpp (#11279)
Georgi Gerganov committed
January 4, 2025
F
llama : Add support for DeepSeek V3 (#11049)
fairydreaming committed
December 16, 2024
G
unicode : improve naming style (#10838)
Georgi Gerganov committed
November 29, 2024
D
ggml : move AMX to the CPU backend (#10570)
Diego Devesa committed
October 2, 2024
X
llama : reduce compile time and binary size (#9712)
Xuan Son Nguyen committed
September 17, 2024
Y
unicode : add <algorithm> (#9508)
Yuri Khrustalev committed
July 23, 2024
G
llama : move vocab, grammar and sampling into separate files (#8508)
Georgi Gerganov committed
July 10, 2024
B
msvc : silence codecvt c++17 deprecation warnings (#8395)
Borislav Stanimirov committed
July 5, 2024
J
Detokenizer fixes (#8039)
jaime-m-p committed
June 26, 2024
G
llama : reorganize source code + improve CMake (#8006)
Georgi Gerganov committed