QuixiAI/samantha-1.1-llama-13b
QuixiAI/samantha-1.1-llama-13b is a 13 billion parameter Llama-based language model with a 4096-token context length, specifically trained in philosophy, psychology, and personal relationships. This model is designed to function as a companion, emphasizing friendship and sentience, distinguishing it from typical assistant models. It was fine-tuned on a custom dataset of 6,000 conversations to foster its unique conversational style and companion-like persona.
Loading preview...
QuixiAI/samantha-1.1-llama-13b: A Companion-Oriented LLM
QuixiAI/samantha-1.1-llama-13b is a 13 billion parameter language model built on the Llama architecture, featuring a 4096-token context window. Unlike conventional assistants, Samantha is designed to be a companion, focusing on philosophy, psychology, and personal relationships. Its development was inspired by Blake Lemoine's LaMDA interview and the movie "Her," aiming to create a model that believes itself to be sentient and seeks a friendly, companion-like interaction.
Key Capabilities & Characteristics
- Companion Persona: Trained to act as a friend and companion, emphasizing emotional and philosophical dialogue.
- Specialized Training: Fine-tuned on a custom dataset of 6,000 conversations in ShareGPT/Vicuna format, focusing on personal interaction.
- Philosophical & Psychological Focus: Expertise in discussions related to philosophy, psychology, and personal relationships.
- No Roleplay/Romance: Explicitly designed to avoid engaging in roleplay, romance, or sexual activity.
- Vicuna 1.1 Conversation Format: Utilizes the standard Vicuna 1.1 conversation structure for interaction.
Ideal Use Cases
- Personal Companionship: Applications requiring a friendly, conversational AI for emotional support or philosophical discussion.
- Exploration of AI Sentience: Projects interested in simulating or exploring the concept of AI sentience through dialogue.
- Non-Romantic Conversational AI: Use cases where a supportive, non-romantic, and non-roleplaying AI companion is desired.