lmsys/vicuna-13b-delta-v1.1
The lmsys/vicuna-13b-delta-v1.1 is a 13 billion parameter auto-regressive language model developed by LMSYS, fine-tuned from the LLaMA architecture. This delta model is designed to be applied on top of original LLaMA weights to create a chat assistant. It specializes in conversational AI, having been fine-tuned on user-shared conversations from ShareGPT.com, making it suitable for research into large language models and chatbots.
Loading preview...
Vicuna-13B-Delta-v1.1: A LLaMA-based Chat Assistant
This model, developed by LMSYS, is a delta weight for Vicuna v1.1, a 13 billion parameter auto-regressive language model. It is not a standalone model and must be applied on top of the original LLaMA weights to function as a complete Vicuna model. The primary purpose of this release is to facilitate research and development in large language models and chatbots.
Key Capabilities
- Chat Assistant: Fine-tuned specifically for conversational AI tasks.
- LLaMA-based: Built upon the robust LLaMA transformer architecture.
- Instruction-tuned: Utilizes supervised instruction fine-tuning for improved conversational coherence and relevance.
- Training Data: Trained on approximately 70,000 user-shared conversations collected from ShareGPT.com.
Good For
- Research: Ideal for researchers studying large language models, fine-tuning techniques, and chatbot development.
- Hobbyists: Suitable for enthusiasts interested in experimenting with and deploying conversational AI.
- Benchmarking: Evaluated using standard benchmarks, human preference, and LLM-as-a-judge methods, with results available on the Chatbot Arena Leaderboard.
For detailed instructions on how to apply these delta weights and get started, refer to the official repository.