Changelog
v1.2.1 (Oct. 23, 2024)
-
Added support for Llama 3.1 8B.
(llama.cpp runtime updated to b3853)
-
Added support for LoRA.
To use LoRA, specify the file path and strength in the Model Setting.
v1.1.1 (Jul. 1, 2024)
- Changed the default context length from 2048 to -1 (the context length the model trained)
- Added workaround for error when saving states of Gemma 2 model
v1.1.0 (Jun. 30, 2024)
-
Added support for Gemma 2.
(llama.cpp runtime updated to b3262)
v1.0.0 (Jun. 11, 2024)
- First release.