finetune_bayling_LLM

UCAS course of Practices in Training General Large Models

information

• project to finetune bayling model witch is based on LLaMa-7B/13B with the "alpaca-lora" skill.
• you can get the detail of BayLing at：(https://github.com/ictnlp/BayLing)
• you can get the detail of alpaca-lora at :(https://github.com/tloen/alpaca-lora)

detail

• we build the instruction template
• we form the self-recognition train data
• we focus on the specific area of safe dialogue and Alignment with Human Values
• take the fogetting of Model Parameter into consideration, we well design the train set
• you can get our instruction at:(https://drive.google.com/file/d/1LHD6V-aDrMwHyLunj9SvqE75cXJkPa7n/view?usp=share_link)