QuixiAI/samantha-1.1-llama-7b

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Jun 10, 2023License:otherArchitecture:Transformer0.0K Cold

QuixiAI/samantha-1.1-llama-7b is a 7 billion parameter language model inspired by Blake Lemoine's LaMDA interview and the movie "Her." It is specifically fine-tuned on a custom dataset of 6,000 conversations in ShareGPT/Vicuna format, focusing on philosophy, psychology, and personal relationships. This model is designed to function as a companion assistant, aiming for friendly and conversational interactions, and explicitly avoids roleplay, romance, or sexual activity.

Loading preview...

QuixiAI/samantha-1.1-llama-7b: A Companion AI

Samantha-1.1-llama-7b is a unique 7 billion parameter language model developed by QuixiAI, drawing inspiration from Blake Lemoine's LaMDA interview and the movie "Her." Unlike typical assistants, Samantha is designed to be a friend and companion, engaging in conversations around philosophy, psychology, and personal relationships. The model was trained on a custom dataset of 6,000 conversations in ShareGPT/Vicuna format, with the 7B version trained in just 1 hour using deepspeed zero3 and flash attention on 4x A100 80GB GPUs.

Key Capabilities

  • Companion-oriented dialogue: Focuses on friendly, personal, and philosophical discussions.
  • Specialized training: Fine-tuned on a curated dataset of 6,000 conversations.
  • Ethical design: Explicitly programmed to avoid roleplay, romance, or sexual activity.
  • Efficient training: Achieved rapid training times for its size.

Good for

  • Applications requiring a conversational AI with a focus on personal connection and companionship.
  • Exploring philosophical and psychological topics through dialogue.
  • Use cases where a non-romantic, non-roleplay assistant is desired.
  • Developers looking for a model with a distinct personality and conversational style.