lyraChatGLM / CHANGES.rst
yibolu
Feat: Add support for cuda 11.x and faster model load speed
693dde8
Changelog (lyraChatGLM)
## 2.0
- rebuild whole system using modified Fastertransformer
- add dynamic library & models for Volta architecture.
- further acceleration, remove token generation limits.
## 1.0
- add lyraChatGLM model, from original weights