Migrate to the latest version of llama.cpp APIs and support for DeepSeek models#1912

Closed

jaepil wants to merge 12 commits intoabetlen:mainfrom

Commits on Jan 21, 2025

Submodule vendor/llama.cpp f7cd133..6171c9d
JamePeng
committed
786fb42
Copy full SHA for 786fb42
Update llama_cpp: Sync LLAMA_API names with llama.cpp mainline. Needs more testing
JamePeng
committed
ff3c9e1
Copy full SHA for ff3c9e1
fix the llama_model_load_from_splits function name
JamePeng
committed
6eb78ec
Copy full SHA for 6eb78ec
correct the llama_tokenize param0 from self.model to self.vocab
JamePeng
committed
1e09d92
Copy full SHA for 1e09d92
fixe more params
JamePeng
committed
60e6b15
Copy full SHA for 60e6b15
fix typo
JamePeng
committed
d1dbd0c
Copy full SHA for d1dbd0c

Commits on Jan 26, 2025

Correct some typo
JamePeng
committed
84bcb2c
Copy full SHA for 84bcb2c
correct llama_chat_apply_template function params
JamePeng
committed
7ddf097
Copy full SHA for 7ddf097
correct var type
JamePeng
committed
4dc2609
Copy full SHA for 4dc2609
Move self._vocab to llama.py
JamePeng
committed
3ffc680
Copy full SHA for 3ffc680

Commits on Jan 27, 2025

Merge pull request #1 from JamePeng/main
Show description for 660955b
jaepil
authored
660955b
Copy full SHA for 660955b
Migrate to the latest version of llama.cpp
jaepil
committed
69c9786
Copy full SHA for 69c9786