Nitral-AI/Poppy_Porpoise-1.0-L3-8B
Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kPublished:May 25, 2024License:otherArchitecture:Transformer0.0K Warm

Nitral-AI/Poppy_Porpoise-1.0-L3-8B is an AI roleplay assistant based on the Llama 3 8B model, developed by Nitral-AI. This model specializes in crafting narrative experiences, tailoring adventures to user preferences with its advanced language capabilities. It also supports vision functionality when used with Koboldcpp and a specified mmproj file. The model achieves an average score of 69.24 on the OpenLLM Leaderboard, demonstrating capabilities across reasoning, common sense, and language understanding tasks.

Loading preview...

Poppy Porpoise 1.0-L3-8B: An AI Roleplay Assistant

Nitral-AI's Poppy_Porpoise-1.0-L3-8B is an instruction-tuned model built upon the Llama 3 8B architecture, specifically designed to function as an AI roleplay assistant. Its core strength lies in generating interactive and engaging narrative experiences, adapting to individual user preferences.

Key Capabilities & Features

  • Specialized Roleplay: Optimized for crafting detailed and immersive narrative adventures.
  • Vision Functionality: Supports multimodal input when integrated with Koboldcpp and the provided Llava MMProj file, allowing for image understanding within roleplay scenarios.
  • Performance: Achieves a competitive average score of 69.24 on the OpenLLM Leaderboard, with specific scores including:
    • AI2 Reasoning Challenge (25-Shot): 63.40
    • HellaSwag (10-Shot): 82.89
    • MMLU (5-Shot): 68.04
    • TruthfulQA (0-shot): 54.12
    • Winogrande (5-shot): 77.90
    • GSM8k (5-shot): 69.07
  • Accessibility: Pre-quantized versions are available, including 5bpw Exl2 and 8bpw Exl2, alongside GGUF's with Imatrix.

Ideal Use Cases

  • Interactive Storytelling: Developers looking to integrate an AI capable of generating dynamic and personalized narratives.
  • Roleplaying Applications: Building applications that require an AI to maintain character, context, and adapt to user input in a roleplay setting.
  • Multimodal Roleplay: For scenarios where visual context is important, leveraging its vision capabilities with Koboldcpp.