dolphin-the-goat/baby-dolphin-2.0
Baby-Dolphin-2.0 is a 3 billion parameter language model based on the Microsoft Phi-2 architecture, fine-tuned by dolphin-the-goat. It is designed for general language tasks, excelling in question answering, chat, and code generation. The model leverages a Transformer architecture with a 2048 token context length, making it suitable for research into safety challenges and exploring non-restricted small models.
Loading preview...
Model Overview
Baby-Dolphin-2.0 is a 3 billion parameter language model, fine-tuned by dolphin-the-goat, based on the Microsoft Phi-2 architecture. Phi-2 is a Transformer model with 2.7 billion parameters, trained on a diverse dataset including NLP synthetic texts and filtered websites. It demonstrates strong performance in common sense, language understanding, and logical reasoning among models under 13 billion parameters. This model is intended for research into vital safety challenges like reducing toxicity, understanding societal biases, and enhancing controllability.
Key Capabilities
- Question Answering (QA): Optimized for direct questions and instruction-based QA formats.
- Chat: Capable of engaging in multi-turn conversational interactions.
- Code Generation: Proficient in generating code, particularly in Python, with common packages.
- Research Focus: Provides a non-restricted small model for exploring safety challenges in AI.
Good For
- AI Safety Research: Ideal for researchers investigating toxicity, bias, and controllability in language models.
- Prototyping QA Systems: Suitable for developing and testing question-answering applications.
- Chatbot Development: Useful for building and experimenting with conversational agents.
- Code Assistant Tools: Can serve as a base for tools that assist with Python code generation and completion.