dphn/dolphin-2.9.1-llama-3-8b
Dolphin 2.9.1 Llama 3 8b is an 8 billion parameter instruction-tuned causal language model developed by Eric Hartford, Lucas Atkins, Fernando Fernandes, and Cognitive Computations. Based on Meta's Llama 3 architecture, it features an 8192 token context length and is fine-tuned for instruction following, conversation, coding, and initial agentic abilities. This model is uncensored and designed for high compliance with requests, making it suitable for developers who require an unfiltered base for custom alignment layers.
Loading preview...
Dolphin 2.9.1 Llama 3 8b Overview
Dolphin 2.9.1 Llama 3 8b is an 8 billion parameter instruction-tuned model developed by Eric Hartford, Lucas Atkins, Fernando Fernandes, and Cognitive Computations. It is a retrained version of Dolphin 2.9, specifically addressing behavioral issues found in the initial dataset by removing 'Systemchat' and 'Ultrachat' to improve system prompt reliance and generation length. The model is based on Meta's Llama 3-8b and utilizes a ChatML prompt template.
Key Capabilities
- Instruction Following: Designed for high compliance with user requests.
- Conversational Skills: Enhanced for natural dialogue generation.
- Coding Abilities: Includes instruction and generation capabilities for code.
- Agentic Features: Possesses initial agentic abilities and supports function calling.
- Uncensored Output: Provides unfiltered responses, allowing developers to implement custom alignment layers.
Good For
- Custom AI Assistants: Ideal as a base for applications requiring highly compliant and unfiltered responses.
- Coding and Development Tools: Suitable for tasks involving code generation and understanding.
- Research and Experimentation: Useful for exploring model behavior without inherent alignment biases.