EmoLLM is a large model project used in the field of mental health. It aims to support user understanding and help users provide mental health counseling by fine-tuning instructions on the large language model (LLM).
- Help users understand and manage emotions
- Improve behavioral patterns and coping strategies
- Provide mental health assessments and interventions
Data sets and domain models
EmoLLM has open-source its dataset, and officials have trained different domain models based on different datasets, such as role-playing, elderly mother psychological counselor, and father boyfriend psychological counselor. The official introduction is as follows:
Datasets are divided into two types by use: General and Role-play
Data is divided into two types by format: QA and Conversation
Data summary: General (6 datasets);Role-play (5 datasets)
Data set type:
General: General dataset, including general content such as psychological knowledge and psychological counseling techniques
Role-play: A role-playing dataset that contains data on specific characters ‘dialogue styles and other content
Taking the role-playing dataset as an example, from my personal analysis, there are two points worth paying attention to:
Modified data format. Transform the domain data format based on the standard sharegpt format or alpaca format for training large models.
Design of prompt words. Design different prompt words for different business scenarios. This is also the meaning and role of the prompt word project.
If you want to learn more, you can click on the link below the video.
Thank you for watching this video. If you like it, please subscribe and like it. thank
Github：https://github.com/SmartFlowAI/EmoLLM
Oil tubing: