TryMore/TryMoreGPT-delta-13b
TryMore/TryMoreGPT-delta-13b is a 13 billion parameter delta model developed by Chuanmo Research Institute, designed to be applied to original LLaMA weights to obtain the full TryMoreGPT model. Utilizing the Vicuna training framework, it has been instruction fine-tuned on a diverse dataset including shareGPT, Alpaca Chinese-English, and COIG universal values and code writing datasets. This model demonstrates competitive performance in Chinese language tasks compared to the original Vicuna and other Chinese chatbots, making it suitable for chat applications requiring strong Chinese language capabilities.
Loading preview...
TryMoreGPT-delta-13b Overview
TryMoreGPT-delta-13b is a 13 billion parameter delta model developed by Chuanmo Research Institute. It is not a standalone model but rather a set of weights designed to be applied to the original LLaMA base model to construct the full TryMoreGPT. This approach allows for efficient distribution and application of the fine-tuned weights.
Key Capabilities
- Instruction Fine-tuning: The model has undergone instruction fine-tuning using the Vicuna training framework.
- Diverse Training Data: Training involved a combination of datasets including shareGPT, Alpaca Chinese-English, and specific universal values and code writing datasets from COIG.
- Enhanced Chinese Performance: TryMoreGPT-13b exhibits strong performance in Chinese language tasks, offering a competitive alternative to the original Vicuna and other Chinese-focused chatbots.
Good For
- Chinese Chatbot Development: Ideal for building chat applications that require robust Chinese language understanding and generation.
- Instruction Following: Its instruction fine-tuning makes it suitable for tasks requiring adherence to specific prompts and commands.
- Research and Development: Provides a foundation for further experimentation and development in large language models, particularly for those interested in its unique delta model application method.