dphn/dolphin-2.9-llama3-8b

Dolphin 2.9 Llama 3 8b Overview

Dolphin 2.9 Llama 3 8b is an 8 billion parameter, full-weight fine-tuned language model built upon Meta's Llama-3-8B. Developed by Eric Hartford, Lucas Atkins, and Fernando Fernandes, and Cognitive Computations, this model was trained using a 4k sequence length on a base model with an 8k context. A key characteristic is its uncensored nature, with the dataset filtered to remove alignment and bias, resulting in a highly compliant model. Users are advised to implement their own alignment layers for responsible deployment.

Key Capabilities

• Instruction Following: Excels at understanding and executing complex instructions.
• Conversational Skills: Capable of engaging in natural and coherent dialogues.
• Coding Proficiency: Possesses strong coding abilities, enhanced by specific coding datasets.
• Agentic Abilities & Function Calling: Includes initial capabilities for agentic workflows and supports function calling, enabling more dynamic interactions.
• Uncensored Output: Provides highly compliant responses, free from inherent alignment biases.

Good For

• Customizable AI Assistants: Ideal for developers who require a highly compliant base model to implement their own safety and alignment layers.
• Coding and Development Tools: Suitable for applications requiring robust code generation, translation, and feedback.
• Research and Experimentation: Useful for exploring the behavior of uncensored models and developing novel AI applications without built-in restrictions.
• Complex Task Automation: Benefits from agentic abilities and function calling for automating multi-step processes.